1. Introduction
Paragraph 1 of Article 1 of the Agreement on Subsidies and Countervailing Measures (SCM Agreement) defines a subsidy as a ‘financial contribution by government’ that confers a benefit. A ‘benefit’ consists of something better than the market would provide.Footnote 1
The governmental practices that would constitute a ‘financial contribution’ are set out in some detail in the first three of four subparagraphs of paragraph 1(a) of Article 1. These include grants, loans, equity infusions, and tax incentives. Subparagraph (iv) is an anticircumvention provision encompassing the financial contributions detailed in the first three subparagraphs when those contributions are made by a private body that a government ‘entrusts or directs’ to make them.
Paragraph 2 of Article 1 further provides that countervailing duties may be imposed only on subsidies that are ‘specific’ to an enterprise or industry, or group of enterprises or industries. Thus a subsidy is the provision of a financial contribution that confers a benefit to a specific enterprise or industry or group of enterprises or industries.
The scope, reach and interpretation of these provisions were the subject of the Appellate Body's June 2005 Report on US – Countervailing Duty Investigation on DRAMS, a case that stemmed from the imposition by the United States of countervailing duties in the amount of 44.29% on imports of DRAMS, a type of semiconductor, from the financially troubled Korean producer, Hynix. The US authorities found that the Government of Korea (GOK) had entrusted or directed private Korean financial institutions to provide beneficial financial contributions, as defined in Article 1 of the SCM Agreement, to Hynix. The central issues in that Report concerned whether the Panel correctly interpreted the ‘entrusts or directs’ language of subparagraph (iv) in concluding that the countervailing duty measure imposed by the US was inconsistent with the requirements of that subparagraph, and whether the Panel properly reviewed the evidence on which the countervailing duty was based.
2. Procedural facts
The WTO dispute began when Korea first asked for consultations with the United States and then requested the establishment of a panel to examine the consistency of the countervailing duty order on DRAMS with the requirements of the SCM Agreement. The Panel found that the duties were inconsistent with both Articles 1 and 2 of the Agreement, the United States appealed, and Korea filed its own appeal on a procedural issue. The Appellate Body modified the Panel's Report by largely reversing its findings.
3. Procedural issues
Rare is the dispute that goes through the WTO process without procedural arguments, and almost as rare is the dispute in which procedural arguments make a difference in the outcome. The DRAMS case, in which the Appellate Body considered two procedural claims – and quickly disposed of them – was no exception. The first procedural claim was by the US, which argued that the Panel had erroneously concluded that Korea's original request for consultations was adequate. The second was the claim by Korea that the US notice of appeal with regard to the consultation request was defective. Since Korea's claim went to the jurisdiction of the Appellate Body even to hear the US claim of Panel error with regard to the consultation request, Korea's procedural appeal was considered first.
In its Notice of Appeal, the United States had said only that it sought review of the Panel's conclusion that ‘Korea's consultation request provides a sufficient indication of the legal basis of the complaint within the meaning of Article 4.4 of the DSU’.Footnote 2 The request, therefore, identified both the issue (the Panel's conclusion with regard to the consultation request) and the legal provision (Article 4.4 of the DSU) whose requirements the request allegedly failed to meet. This, the Appellate Body held, was sufficient, and Korea's jurisdictional appeal was denied.
The Appellate Body then considered the US argument regarding the alleged deficiency of Korea's consultation request on its merits. The consultation request dispute stemmed from the fact that, early in the process, Korea had filed an addendum to its original request. The initial request sought consultations only on the preliminary and final subsidy determinations of the US Department of Commerce and on the preliminary injury determination by the US International Trade Commission. The addendum sought additional consultations on the final injury determination and on the countervailing duty order – the ‘measure’ – itself. Two sessions of consultations in fact were held, one on the initial request and the other on the addendum.
Nonetheless, the US argued that, because the final injury determination and the countervailing order were not themselves part of the initial request, Korea was wrong to include them in its request for the establishment of a panel, and the Panel was wrong to consider them. The Appellate Body disagreed and upheld the Panel. It then moved on to the first key substantive legal question: the meaning of the words ‘entrusts or directs’ in subparagraph (iv) of Article 1.1(a) of the SCM Agreement.
4. The meaning of ‘entrusts or directs’
We are trapped by our language. There is perhaps no better illustration of this than the effort of legal tribunals to interpret legal language, such as treaty language. Words are defined in terms of other words, a process that, in the end, is circular. But, as we proceed around the circle, emphasis, nuance, even meaning can change. Those changes in emphasis, nuance, and meaning often create difficulties when efforts are made to explain what a legal term requires.
The task of the Panel in DRAMS was to interpret the words ‘entrusts or directs’ in subparagraph (iv). The same task had faced the Panel in US – Export Restraints, and the DRAMS Panel simply followed the Export Restraints Panel's interpretation – an interpretation that was not appealed.Footnote 3DRAMS, therefore, was the Appellate Body's first brush with the ‘entrusts or directs’ requirement, and it disagreed with both Panels.
Following the reasoning in Export Restraints, the Panel held that ‘entrusts’ contains the notion of ‘delegation’, while ‘directs’ contains the notion of ‘command’.Footnote 4 The Panel, the Appellate Body said in disagreement, ‘effectively replaced the terms “entrusts” and “directs” with two other terms, “delegation” and “command”, whose scope it did not define, and went no further in clarifying the meaning of any of these terms’.Footnote 5 ‘Entrusts or directs’, it said, does not have the same meaning as ‘delegates or commands’.Footnote 6
‘Delegation’, the Appellate Body said, usually involves a formal action, while ‘entrusts’ connotes giving responsibility to someone.Footnote 7 Thus, it found too narrow an interpretation of ‘entrusts’ to mean simply ‘delegation’.Footnote 8 Similarly, while governments can exercise authority by command, they are likely to have other, more subtle, means at their disposal. ‘Thus, an interpretation of the term “directs” that is limited to acts of “command” is also too narrow.’Footnote 9
The Appellate Body acknowledged, however, that identification of the kinds of government actions that could constitute entrustment or direction is difficult in the abstract. It noted in this regard that Korea admitted that ‘in some circumstances “guidance” by a government can constitute direction’.Footnote 10 In the end, however, the Appellate Body did not rest its decision on abstract definitions. Instead, it said that a determination of entrustment or direction hinges on the specific facts of a case.Footnote 11 The facts of the DRAMS case, in particular the method of their evaluation, raised one of the more interesting issues in the Report: what is the correct way to evaluate circumstantial evidence?
5. Evidentiary issues
The evidentiary findings
Circumstantial evidence is indirect evidence. It does not go directly to the question at issue, but permits the answer to that question to be inferred by its consistency with a possible answer. The DRAMS case was all about circumstantial evidence.
In concluding that Korea had conferred a subsidy on Hynix within the meaning of Article 1 of the SCM Agreement, the US found that four financial contributions were made by both public and private creditors who were entrusted or directed to do so by the GOK. The contributions included an 800-billion-won syndicated loan, a ‘Fast Track Programme’ by the Korea Development Bank, and two restructuring packages. The objective of these contributions, according to the US, ‘was the complete financial restructuring of Hynix in order to maintain the company as an ongoing concern’.Footnote 12
Some of the creditors were public bodies, others were banks that were owned or controlled by the GOK, while the remainder were private entities in which the Government had only a small interest or none at all. Korea conceded that financial contributions by public creditors were within the scope of Article 1 (although they did not agree that these met the specificity requirements of Article 2, which is discussed below). The core of the dispute revolved around whether the US finding that the GOK entrusted or directed the other creditors to make the financial contributions was supported by the evidence.
The Panel considered four items of circumstantial evidence that the US relied upon in finding a financial contribution: (1) Korea's policy of preventing the failure of Hynix; (2) its control of Hynix's creditors; (3) its coercion of those creditors; and (4) a prospectus filed by two Korean banks with the US Securities and Exchange Commission. The Panel did conclude that ‘an objective and impartial investigating authority could properly have found that the GOK had a policy to save Hynix’.Footnote 13 However, while this policy could explain the participation of the public body creditors in the financial contributions, the Panel found that it was not sufficient to attribute to the GOK the participation of the other creditors. The ‘existence of a GOK policy to save Hynix in and of itself could not properly be treated as evidence of government direction’ of the non-public creditors, the Panel said.Footnote 14
In evaluating the remaining three items of circumstantial evidence, the Panel found that, while the US had established that the GOK had regulatory control over the private creditors (including the government-owned or controlled banks), it had ‘not properly established that the GOK actually exercised such influence or regulatory authority’.Footnote 15 Similarly, although the US demonstrated that one creditor had been coerced, it had not demonstrated to the Panel's satisfaction that others had been coerced.Footnote 16 Finally, it concluded that the prospectus filed with the US Securities and Exchange Commission did not establish that the filing institutions had been entrusted or directed to support Hynix.Footnote 17
The Panel also rejected the US position that the four financial contributions at issue should be considered to be part of a single program, rather than four separate programs. Thus, in the Panel's words, ‘although the [US] established that the GOK had a policy to save Hynix, and that the GOK had a certain capacity to influence [the non-public body] creditors … the [US] did not properly demonstrate that the GOK availed itself of that capacity to entrust or direct all [non-public body] creditors to participate in all four of the financial contributions at issue’.Footnote 18
Analysis of circumstantial evidence
A single piece of circumstantial evidence rarely, if ever, would be enough to justify a legal conclusion. Circumstantial evidence cases are built on the cumulation of evidence. The question before the Appellate Body was whether the Panel's approach to analyzing more than one piece of circumstantial evidence was correct. The Appellate Body concluded that the Panel's approach was in error.
It began its analysis with the observation that, although the Panel stated that it would follow the same approach that the US employed – that is, that it would base its conclusion on the totality of the evidence – the Panel actually did the reverse.Footnote 19 ‘[T]he Panel often appeared to examine whether each piece of evidence, viewed in isolation, demonstrated entrustment or direction.’Footnote 20 As an example, the Appellate Body noted that the US, in its countervailing-duty decision, had concluded that a GOK financial supervisory agency had increased credit limits for bank loans to single borrowers. The Panel concluded that, even though the US might have been correct in arguing that some of the creditors would not have been able to participate in a syndicated loan to Hynix without the waiver, the US could not ‘properly have inferred from this that the creditors were instructed or directed to participate in the syndicated loan’.Footnote 21
‘We do not read the [US] to have inferred solely from the waiver of loan limits that entrustment or direction had taken place’, the Appellate Body said. ‘Nor do we consider that the [US’s] reliance on this evidence suggests that “there is government entrustment or direction every time that a loan limit waiver is provided”, for this would follow only if the [US] had based its finding of entrustment or direction exclusively on the waivers of loan limits.'Footnote 22 The waiver of loan limits had to be looked at in light of all of the actions the GOK took that had a bearing on its policy to save Hynix.
It is a legal truism that factual findings are to be made on the basis of an examination of the totality of the evidence, particularly circumstantial evidence. Even the Panel acknowledged this. Yet legal tribunals have seldom, if ever, explained why this is the case, and the Appellate Body's Report in DRAMS is no exception.
Despite its protestations to the contrary, the Panel, as the Appellate Body said, seemed to view each piece of evidence separately, in isolation, and if that piece – by itself – was not enough to establish entrustment or direction, it was put aside and the next piece was examined in the same way. The Panel correctly did not consider a process of addition, whereby evidence that fell short of the required level by itself would be added with similar pieces of evidence to reach a cumulative total. This apparently was the approach of Korea, which argued that it is ‘a mathematical truism that no matter how many zeros and negative numbers one adds together, the sum can never be a positive number’.Footnote 23
Of course, adding zeros and negative numbers will not reach a positive number, but Korea's argument ignores the fact that the Panel did find some positive numbers. It found, for example, that Korea had a policy of preventing the failure of Hynix and that at least one non-public creditor had been coerced into participating in the rescue operation. But more importantly, if pieces of evidence are given a quantitative value, it is error to add them. To do so is to commit the error of the student who, after taking two examinations and receiving a 50% score in each, tells his parents that his score for the two exams was 100%.
When legal tribunals make factual determinations based on the totality of several pieces of circumstantial evidence, they are really engaged in an exercise in probability. There is one chance in two that a flipped coin will come up heads, but only one in four that it will come up heads twice in a row, one in eight that it will do so three times in a row, one in 16 that it will do so four times in a row, and so forth. If, after a single flip, a coin comes up heads, there is no rational basis for inferring that the coin has been altered in some way. But if it comes up heads a number of times in a row, an inference that the coin might have been altered would be reasonable.
Although evaluation of circumstantial evidence is based on an informal probability analysis, the process differs from simple coin-flipping probability because individual pieces of evidence, unlike coin flips, also have a qualitative value. One coin flip is no more important than another, but the importance and reliability of pieces of evidence can vary greatly. These factors can make an evaluation of the totality of the evidence difficult; nonetheless it is clear, as the Appellate Body concluded, that all of the evidence has to be looked at together and a judgment made as to how much the coincidence of the individual pieces supports a conclusion that none of them, alone, could support.
Other evidentiary issues
1. Admissibility
In its submissions to the Panel, the United States referred to evidence that was part of the record of its countervailing duty investigation, but had not been cited in its written decision to impose the duties. The Panel refused to consider this evidence on the ground that it amounted to an ex post rationalization.Footnote 24 The Appellate Body disagreed.
While Members may not attempt to defend countervailing duty decisions on the basis of evidence not contained in the investigative record, the evidence at issue concededly was contained in that record. And, while it is true that Article 22.5 of the SCM Agreement requires a final countervailing duty determination to contain ‘all relevant information on the matters of fact and law and reasons which have led to the imposition’ of duties, it does not require Members ‘to cite or discuss every piece of supporting record evidence for each fact in the final determination’.Footnote 25 In referring to the evidence at issue, the Appellate Body reasoned, the US was not advancing new reasons to support its decision, but was simply adding to the support of its reasons by referring to further record evidence. Accordingly, the Panel erred in refusing to consider this evidence.Footnote 26
2. Non-record evidence
A claim by the US that the Panel itself had relied on non-record evidence was rejected in those terms by the Appellate Body, but, nonetheless, the Panel's conclusion with regard to the evidence was found to be erroneous.
Korea argued that, despite its ostensible control over the creditors of Hynix, its laws provided that any creditor not satisfied with proposed restructuring terms acceptable to a majority of the creditors could exercise an option to be bought out by the creditors who found the terms acceptable. If a buyout price could not be agreed to, the amount could be settled through mediation.Footnote 27
A paragraph in Hynix's audit report, which was part of the record of the US investigation, stated that three creditors raised an objection to the terms of the proposed restructuring, but the paragraph gave no further information. The Panel concluded from this paragraph that the US should have understood that the three creditors had exercised an independent right to object and be bought out according to terms set through the mediation process. The Appellate Body, however, found that such a factual inference was not reasonably suggested by the text of the paragraph in question.Footnote 28 In fact, according to the Appellate Body, the US during its investigation had explicitly asked Hynix whether any creditors had exercised their buyout rights and received no indication from Hynix that this had occurred.Footnote 29 Accordingly, the Appellate Body found that, in reaching its conclusion on this issue, ‘the Panel failed to “make an objective assessment of the matter before it” as required by Article 11 of the DSU’.Footnote 30
3. Standard of review
The US appeal did not challenge the Panel's articulation of the proper standard for reviewing the countervailing duty determination; rather, it challenged the Panel's application of that standard. The Appellate Body agreed with the US.
The only standard of review contained in the DSU is set forth in Article 11, which specifies that a panel ‘should make an objective assessment of the matter before it, including an objective assessment of the facts of the case and the applicability of and conformity with the relevant covered agreements’. This general language was interpreted by the Appellate Body to mean that the ‘objective assessment’ called for by Article 11 requires a panel reviewing a Member's subsidy determination to ‘be informed by an examination of whether the agency provided a reasoned and adequate explanation as to (i) how the evidence on the record supported its factual findings; and (ii) how those factual findings supported the overall subsidy determination’.Footnote 31 This means that a panel may not conduct its own de novo investigation or substitute its judgment for that of the Member.Footnote 32 A panel, the Appellate Body pointed out, acts ‘as a reviewer of agency action, rather than as [an] initial trier of fact’.Footnote 33
In reviewing an individual piece of evidence, a panel should focus on such items as its accuracy and whether it may reasonably be relied upon to support the inference drawn by the investigative agency.Footnote 34 In this dispute, however, the Panel not only failed to examine the evidence in its totality, it also ‘examined whether certain pieces of evidence were sufficient to establish certain conclusions that the [US] did not seek to draw, at least solely on the basis of those pieces of evidence’.Footnote 35 Thus, in the view of the Appellate Body, the Panel failed to review the US determination, but reached ‘its own view of whether entrustment or direction existed in this case’.Footnote 36 This de novo review of the evidence was contrary to the proper standard of review and the requirements of Article 11.Footnote 37
6. Benefit
In its countervailing-duty investigation, the US rejected loan terms offered by private creditors of Hynix as market benchmarks, based on its finding that the creditors were entrusted or directed to offer those terms. The Panel concluded that, since the US erroneously found entrustment or direction with regard to these creditors, its rejection of them as market benchmarks was also erroneous. Since the Appellate Body reversed the Panel's findings on entrustment or direction, it found ‘no basis for us to uphold the Panel's finding on benefit’. Accordingly, it reversed that determination.Footnote 38
7. Specificity
The countervailing duty determination found that the financial contributions from all of the creditors were ‘specific’ within the meaning of Article 1.2 of the SCM Agreement because the program was directed toward a single enterprise, Hynix.Footnote 39 Because the Panel found no entrustment or direction with regard to the non-public creditors, however, it also found that the US finding of specificity with regard to these creditors was improper.Footnote 40 Following the logic of its reversal of the Panel's entrustment or direction, the Appellate Body also reversed the Panel's findings as to specificity with regard to the non-public creditors.Footnote 41
8. No ‘completing the analysis’
As is well known, the Appellate Body does not possess remand authority.Footnote 42 This means that when it reverses a panel's decision, the dispute usually is effectively ended, even though the nature of its reversal may leave crucial questions unanswered and may deny a complainant a decision on its claims.
A typical instance of this phenomenon occurs when a complainant asserts that the responding Member's measure is inconsistent with two or more of its WTO obligations. A panel may find in the complainant's favor in one instance, and – practicing judicial economy – conclude that it is unnecessary for it to decide the other. If the Appellate Body reverses the Panel's decision of a violation in the single decided issue, the second claim may simply be left undecided.
To avoid this situation whenever possible, the Appellate Body has developed a practice of ‘completing the analysis’. When the factual findings of the Panel may provide an adequate basis for it to do so, the Appellate Body itself will supply the missing legal analysis on the undecided claim on the basis of those facts. But in some instances, the crucial facts have not been found by the Panel, and the record does not permit the Appellate Body to ‘complete the analysis’. DRAMS was such an instance, even though it was not an example of the more typical two-claim case. In DRAMS, because the Panel employed the wrong methodology, it never decided some of the factual issues that it would have to have decided in order to employ the proper methodology.
The Appellate Body in DRAMS only reversed the finding of the Panel that the US was wrong in concluding that Korea entrusted or directed private creditors to provide benefit-conferring financial contributions to Hynix. However, the Appellate Body did not decide the opposite. It did not decide that the US was correct in finding that Korea entrusted or directed the private creditors to provide those contributions. Whether the countervailing-duty order could have withstood a proper approach to reviewing the evidence is a question left unanswered. Because the evidentiary record was so sparse on issues that the Appellate Body felt it would need to examine to complete this analysis, it declined to do so and limited its conclusion on entrustment or direction, as well as benefit and specificity, to its examination of the Panel's legal conclusions, nothing more.Footnote 43
9. Conclusion
The Appellate Body's legal conclusions in the DRAMS case seem sound. The Panel's consideration of the evidence was faulty, and this faulty consideration led to its failure to decide evidentiary issues that would have been relevant to a proper analysis. Consequently, after reversing the Panel on many points, the Appellate Body was unable to complete the analysis, could not remand the matter to the Panel for further proceedings, and, therefore, made no recommendation to the Dispute Settlement Body. The result was that, nearly two years after it invoked the dispute-settlement process, Korea was left without a decision on its claim that the US imposition of countervailing duties on exports of DRAMS from Hynix was contrary to the SCM Agreement. This hardly seems to square with the statement in Article 3.3 of the DSU:
The prompt settlement of situations in which a Member considers that any benefits accruing to it directly or indirectly under the covered agreements are being impaired by measures taken by another Member is essential to the effective functioning of the WTO and the maintenance of a proper balance between the rights and obligations of Members.
It seems clear that the dispute-settlement process would be improved by granting remand authority to the Appellate Body.
