The paper by Vermulst and Prusa discussing decisions by WTO Panels and the Appellate Body in respect of two comprehensive challenges of the practice of zeroing in antidumping proceedings explains, in a clear and thorough manner, how this zeroing methodology operates and the effect it has on the determination of dumping margins. Such a proper description of this practice and the effect it has on the determination and calculation of dumping margins is essential for understanding the legal discussion that has been raging at the WTO in respect of zeroing since 1999. Vermulst and Prusa thus provide an important and very useful contribution to the ongoing discussion about the use of zeroing in antidumping investigations, as Members continue to fight over this issue both in dispute settlement, and in the context of the Doha Round negotiations on a new WTO Antidumping Agreement (‘AD Agreement’).
The paper is very rich in terms of explaining the various ways in which zeroing can affect the margins of dumping, and the authors' clear and readable discussion of this technical question of antidumping law deserves a lot of praise. It is not my intention in this Commentary to repeat the excellent analysis of Vermulst and Prusa.Footnote 1 I will rather address a couple of the legal arguments and conclusions of Vermulst and Prusa that merit some additional comments. In addition, I will raise a more systemic question relating to the zeroing case law and examine briefly who is to be held responsible for the fact that this technical antidumping question is still being litigated now, almost ten years after it was first raised in WTO dispute settlement.
1. The law and economics of zeroing – worlds apart?
1.1. Zeroing and fairness
The economic analysis of the practice of zeroing in Vermulst and Prusa is clear: zeroing inflates the margins of dumping and allows an investigating authority to find dumping where no dumping would have been found and no duties would have been imposed otherwise. The conclusions to be drawn, after reading the explanation offered and the examples provided, is undoubtedly that zeroing is unfair and biased as it makes it more likely that dumping will be found to exist. Moreover, as clearly stated by Vermulst and Prusa, it is never the case that zeroing will lead to a lower dumping margin. In other words, zeroing is biased in favor of the petitioner industry and never benefits exporters. It is thus lacking in evenhandedness, and therefore, unfair.
The clear demonstration by Vermulst and Prusa of such lack of fairness contrasts with the legal analysis in which, somewhat surprisingly, they conclude that fairness is a concept that is ‘inherently vague’, suggesting that no finding of violation can be based on such a concept. Vermulst and Prusa state that ‘What is fair in the eyes of an exporter is not necessarily the same as what is fair in the eyes of a domestic producer.’ In so doing, Vermulst and Prusa are echoing the analysis of the Panels in the cases discussed, which, for that same reason, refused to consider that zeroing was unfair or violated the general requirement of Article 2.4 of making a fair comparison between normal value and export price.
In my view, the authors' own well-argued explanation of how the practice of zeroing inflates the margins of dumping and never works to the benefit of exporters is the best evidence of the fact that zeroing is inherently unfair and thus inconsistent with the disciplines of the Antidumping Agreement. The authors' correct economic analysis shows that zeroing is unfair. It seems difficult to accept that a legal analysis would lead to a different conclusion. Vermulst and Prusa seem to suggest that the concept of fairness is too vague and undefined to allow a legal finding to be based on such a concept. However, it is clear that WTO case law in respect of trade-remedy investigations is almost entirely based on similarly ‘vague’ concepts as Panels examine whether a ‘reasonable’ explanation has been provided as to how the facts support the determination made, whether a ‘determination’ was made in the case of reviews, or whether an examination of the injury factors was ‘objective’. In the case of zeroing, the question is not, as suggested by Vermulst and Prusa, whether exporters or the domestic producers consider that the practice of zeroing is unfair or not; but whether the Panel reaches this conclusion. ‘Fairness’ is clearly not a concept that is more vague than ‘reasonableness’ or ‘objectiveness’, which are necessarily employed by Panels all the time.
The main reason why both the Panels and the authors seem to be so keen to avoid the obvious legal conclusion that zeroing is unfair relates to the fact that it arguably would imply a violation of the requirement of Article 2.4 of making a fair comparison between normal value and export price. As the informed reader will know after having read the paper by Vermulst and Prusa, the discussion relating to zeroing started under a subparagraph of this provision, Article 2.4.2 AD Agreement, in which the three methodologies for comparing normal value and export price are set forth. All of the textual arguments made thus far that zeroing is not prohibited by the text of this provision would be rendered moot if, indeed, zeroing is considered as unfair under the generally applicable ‘fair comparison’ requirement of Article 2.4 to which Article 2.4.2 expressly refers. In order to maintain the viability of some of the textual arguments, as developed by certain WTO Members such as the United States and as adopted by a number of Panels under Article 2.4.2, one has to reach the bizarre conclusion that at least from a legal point of view, zeroing is not as ‘inherently unfair’ as the numbers tend to show. However, it seems more reasonable to conclude that what this internal contradiction reveals is that the textual arguments under Article 2.4.2 AD Agreement are simply not that strong. These textual arguments are contradicted by the commonsense requirement that an antidumping investigation be ‘unbiased and objective’ as required by Article 17.6(i) AD Agreement.
This is not to say that I would agree with the Appellate Body that zeroing is a violation of the ‘fair comparison’ requirement of Article 2.4 AD Agreement. Indeed, perhaps the more important comment to make in respect of the Appellate Body's finding in respect of the inconsistency of zeroing with the fair-comparison requirements of Article 2.4 AD Agreement is that this provision only requires an authority to make a fair comparison between normal value and export price but does not set forth a general fairness requirement. Arguably, zeroing has nothing to do with ensuring an apples-to-apples comparison, which is what is actually required by Article 2.4, as the problem is not related to the comparison of normal value with export price, but rather relates to the way the authority deals with the results of such comparisons as it zeroes the negative margins. In my view, Article 2.4 is merely an expression of the general underlying requirement that any antidumping investigation be objective and unbiased and that the laws and regulations relating to the conduct of antidumping investigations are equally conducive to such investigations. Zeroing is clearly at odds with this general requirement, which finds its expression, inter alia, in Article 17.6(i) of the AD Agreement. In sum, whether in law or in economics, zeroing is unfair and therefore inconsistent with the AD Agreement.
1.2. Zeroing and mathematical equivalence
The authors' legal analysis reaches a conclusion different from that reached under their economic analysis in respect of another important aspect of the zeroing debate as well. In their legal analysis, the authors seem to side with the views expressed by some Panels that zeroing cannot be considered as unfair and outlawed in a general manner as this would render inutile the third, exceptional method for dealing with targeted dumping, as set forth in Article 2.4.2 of the AD Agreement. This is the ‘mathematical equivalence’ argument, which posits that without zeroing, the third method (weighted-average normal value to transaction-specific export price) would lead to results that are mathematically equivalent to those obtained under the first method (a weighted-average normal value to weighted-average export-price comparison). Over time, this equivalence argument has become the prime argument to argue that zeroing must be permitted in certain circumstances, and cannot therefore be considered as generally ‘unfair’. The Panels on US–Zeroing (EC) and US–Zeroing (Japan) as well as US–Softwood Lumber V, Article 21.5 argued that the third methodology necessarily permits the use of zeroing, as only through zeroing nondumped export transactions would the result of this third methodology differ from the results reached by applying the first (weighted-average-to-weighted-average) methodology. If this is so, the Panels argued, then the Appellate Body's findings, even though based on terms that could relate to all three methodologies, must be limited to the first methodology, because extending the prohibition of zeroing to the second and third methodology would deprive of all meaning the third methodology. Effective treaty interpretation would argue against such a conclusion, and it cannot be assumed that the Appellate Body wanted to reduce to a nullity a provision of the Agreement. Moreover, according to the Panels, if zeroing is permitted under the third methodology, it cannot be considered to be unfair per se, or inconsistent with the general fair-comparison requirement of Article 2.4 AD Agreement.
The Appellate Body first rejected this mathematical-equivalence argument in its report on US–Softwood Lumber V, Article 21.5.Footnote 2 Interestingly, Vermulst and Prusa equally show that the basic premise of this argument is in error as it is simply not so that the third method becomes meaningless without zeroing. As stated in their paper, the exceptional method can produce different dumping amounts than the two ordinary methods, even without invoking zeroing. Yet, in the legal analysis, the authors keep repeating the argument that the Appellate Body has rendered the entire exceptional method inutile. It is clear that there may be circumstances where the law and the economics lead to different results. However, it seems that in a case like this one, where the legal argument is based on a quasi-economic justification of ‘mathematical equivalence’, the economic analysis that demonstrates the fallacy of this mathematical-equivalence argument must be given due weight. It seems difficult to continue to make this argument in law if the mathematics relied on to support this argument prove to be wrong.
2. Zeroing – merely a matter of transparency?
Another issue dealt with in the cases discussed by Vermulst and Prusa concerned the question whether the ‘zeroing methodology’ as consistently used by the United States can be challenged ‘as such’, independent from its actual application in a particular case. The discussion of whether such a methodology is a ‘measure’ and whether this measure is the line in the computer program used to calculate margins of dumping or rather the methodology in a more general manner is really much ado about nothing. The Appellate Body rightly considered that a methodology which is consistently used is challengeable and does not need to take the form of a law or regulation in order for it to be successfully challenged before the WTO. Vermulst and Prusa agree with the Appellate Body. However, they express the view that the US lost the claims against the zeroing methodology ‘as such’ because the US was ‘again the victim of its own transparency’. In this respect, they refer to another set of cases relating to a number of challenges against the US practice in sunset reviews, as set forth in the US Sunset Policy Bulletin (‘SPB’). In the paper by Vermulst and Prusa, it is suggested that transparency was also a key reason why the US lost several Appellate Body challenges to its sunset-review practice. The sunset case related to a consistent application of a particular methodology for determining a likelihood of recurrence or continuation of dumping in sunset reviews that would justify the extension of the antidumping measure for another five-year period.
It is interesting that the authors raise the Sunset Policy Bulletin cases in this context. Actually, it seems worth mentioning that there is an important difference in approach between the zeroing cases and these sunset cases. In the sunset cases, the US was actually able to rebut the challenges against its Sunset Policy Bulletin, which was never found to be WTO-inconsistent by the Appellate Body. The main reason why the Appellate Body refuses to find against the US in respect of the Sunset Policy Bulletin is because this policy document, with no binding force, contains language that suggests that the authority is not required to act in a manner which would be inconsistent with the requirement of the Antidumping Agreement, i.e. to base a sunset review on facts and not presumptions. The SPB provides for three scenarios that, if demonstrated to be present, are deemed to show that dumping is likely to continue or recur in the future.
The Appellate Body considered that such scenarios may be a useful tool, but their application should be case specific, and they should never be considered as determinative. However, even though the complainants were able to show that the US examines the likelihood of dumping in sunset reviews in 100% of the cases on the basis of the three scenarios of the Sunset Policy Bulletin, and does not examine any other factual evidence, and despite the absence of even one counterexample by the US, the Appellate Body considered that the Sunset Policy Bulletin was not WTO-inconsistent ‘as such’. The Appellate Body found that it had not been demonstrated that the authority considered that it was bound to follow the scenarios of the SPB or that it was precluded from examining other evidence. The 100% consistent application of the Bulletin was rejected by the Appellate Body in United States – Sunset Reviews of Anti-Dumping Measures on Oil Country Tubular Goods from Argentina as ‘mere statistical evidence’.Footnote 3
The difference in approach by the Appellate Body in respect of the zeroing methodology for which no written ‘policy document’ of any sort exists is striking. In respect of zeroing, the 100% consistent application of the methodology was considered determinative to establish the existence of a measure which is, ‘as such’, inconsistent with the WTO Agreements. In other words, it seems that if a WTO Member wants to avoid a finding of inconsistency of its practice, it is better to have some sort of policy document to refer to, as the effect of the mandatory/discretionary distinction is such that any potential for WTO-consistent behavior will allow a country to argue its way out of a 100% inconsistent application of the challenged norms.
Returning to the zeroing cases, it seems too easy to conclude that the US lost these cases because it was the victim of its own transparency. To argue that the US was the victim of its own transparency seems to suggest that transparency is a mitigating circumstance; it suggests that many WTO Members zero, but that the US simply does so openly. This is not correct. The general opposition in the course of the Doha Negotiations to proposals by the US to introduce a provision that expressly sanctions the use of zeroing is proof of the fact that zeroing is very much a US practice. The US was found to have violated the WTO Agreements not because it was too transparent, but because the zeroing methodology is WTO inconsistent. The transparent application of this methodology may well have made it easier to demonstrate the violation, but that is simply a matter of evidence.
3. The zeroing saga – a systemic failure?
Although it is now clearly settled that zeroing is not permitted under the AD Agreement, the road to this conclusion has been particularly thorny. In total, ten casesFootnote 4 and 16 reports so far,Footnote 5 have dealt with the issue of zeroing. The zeroing case law is made up of diverging Panel reports, two dissenting opinions,Footnote 6 a Panel overturning itself,Footnote 7 the Appellate Body changing its reasoning to deal with the issue on several occasions, and includes a number of Panels openly disagreeing with the Appellate Body and refusing to follow its reasoning.Footnote 8 The zeroing saga is still continuingFootnote 9 in spite of what is now a clear and comprehensive set of rulings outlawing zeroing. This begs the question who is to blame for this seemingly endless string of cases dealing with something that is, after all, a very technical aspect of antidumping investigations. While Vermulst and Prusa do not address this question, I would like to offer a couple of reflections in this respect.
In a bizarre way, the cases discussed by Vermulst and Prusa and the other cases relating to the practice of ‘zeroing’ reveal some of the problems in terms of securing predictability and security that affect the Appellate Body's jurisprudence. The zeroing case law started with a Panel adopting a position as close as it gets to a rule of binding precedent (stare decisis) and (for now) ended with Panels for the first time openly refusing to follow the Appellate Body's reasoning and rejecting the Appellate Body's interpretation.
The search for the correct legal basis on which to analyze the practice of zeroing is what characterizes the zeroing case law. It is clear that the Antidumping Agreement does not contain an express prohibition or permission of zeroing. Actually, the term is not mentioned in the Antidumping Agreement. This does not necessarily mean that zeroing is permitted under the Agreement, as became clear from the case law. Still, the problem the Panels and the Appellate Body obviously struggled with was finding the right textual hook to hang the analysis on.
The textual basis for much of the Panels' and the Appellate Body's analysis was Article 2.4.2 AD Agreement dealing with the methods for comparing normal value and export price in order to establish a margin of dumping, and the general requirement to conduct a fair comparison between normal value and export price of Article 2.4 AD Agreement.Footnote 10
The basic problem and the reason for the fact that the zeroing discussion dragged on for so long was the fact that, for too long, the Appellate Body failed to call the thing by its name. It is clear that the practice of zeroing is at odds with the requirement to conduct an objective and unbiased investigation into the existence of dumping.Footnote 11 The requirement of evenhandedness is crucial in conducting an objective examination, and, as the Appellate Body has made clear on numerous occasions, the use of a methodology which makes it more likely that the investigating authority will determine that there is dumping, as is the case with zeroing, is at odds with this obligation.Footnote 12
In an antidumping investigation, there will always be many transactions that will need to be compared, sometimes for different models. So, under any method, the investigating authority will need to calculate an average in the end, whether it is the average of the results of all comparisons of export transactions with domestic transactions, or the average of the results of weighted-average-to-weighted-average comparisons for different models. Averaging precisely implies that positive results will be offsetting negative results. By zeroing, the investigating authority tinkers with the results by disallowing the positive results to fully offset the negative results, thus making it more likely, if not almost inevitable, that a dumping margin will be found. Given this basic problem with zeroing, it is clear that in the absence of an explicit authorization in the text of the AD Agreement that would allow for this type of distorted examination, zeroing has to be considered as prohibited under the AD Agreement. The argument that an explicit prohibition was discussed, but no agreement was reached during the Uruguay Round for the inclusion of such a prohibition, does not convince.Footnote 13 Actually, the fact that zeroing was discussed and no explicit authorization of this obviously biased practice was included in the AD Agreement, only confirms its illegitimate status.
The problem was that the Appellate Body, prior to its report on US–Softwood Lumber V, Article 21.5, had, with one notable exception, shied away from making this basic point that would have ended all further discussion on this matter. Indeed, on one occasion, in its report on US–Corrosion-Resistant Steel Sunset Review, a dispute that only indirectly related to the question of zeroing, the Appellate Body stated in a very straightforward manner that zeroing is inherently unfair and therefore inconsistent with the AD Agreement, irrespective of whether it is applied in original investigations or in reviews:
In EC–Bed Linen, we upheld the finding of the panel that the European Communities acted inconsistently with Article 2.4.2 of the Anti-Dumping Agreement by using a ‘zeroing’ methodology in the anti-dumping investigation at issue in that case. We held that the European Communities' use of this methodology ‘inflated the result from the calculation of the margin of dumping’. We also emphasized that a comparison such as that undertaken by the European Communities in that case is not a ‘fair comparison’ between export price and normal value as required by Articles 2.4 and 2.4.2.
When investigating authorities use a zeroing methodology such as that examined in EC–Bed Linen to calculate a dumping margin, whether in an original investigation or otherwise, that methodology will tend to inflate the margins calculated. Apart from inflating the margins, such a methodology could, in some instances, turn a negative margin of dumping into a positive margin of dumping. As the Panel itself recognized in the present dispute, ‘zeroing … may lead to an affirmative determination that dumping exists where no dumping would have been established in the absence of zeroing’. Thus, the inherent bias in a zeroing methodology of this kind may distort not only the magnitude of a dumping margin, but also a finding of the very existence of dumping.Footnote 14 (footnotes omitted) (emphasis added)
In all of the cases more specifically dealing with zeroing that followed this case, the Appellate Body was looking for a textual hook to prohibit zeroing and failed to outlaw zeroing for what it was, biased and unfair. The textual arguments of the Appellate Body proved to be problematic and became a source of as much discussion as, if not more than, the actual text of the AD Agreement. A number of Panels, as well as some dissenting panelists, all pointed to a couple of technical and textual shortcomings of the Appellate Body's arguments.
The Appellate Body had to move away from the ‘all comparable transactions’ language relied on in EC–Bed Linen to avoid being trapped in a prohibition that related exclusively to the first methodology of Article 2.4.2 AD Agreement. In US–Softwood Lumber V, it thus relied on the term ‘margin of dumping’ as it appeared at the beginning of the opening sentence of Article 2.4.2 AD Agreement. However, its reasoning was still very much linked to Article 2.4.2 AD Agreement. The Panel in US–Zeroing (EC) pointed to the fact that Article 2.4.2 limits its application to the ‘investigation phase’, and on that basis followed the US in its argument that zeroing was not prohibited in the context of administrative reviews. Rather than dealing with this textual, albeit unconvincing argument by the Panel, the Appellate Body again changed the basis for its finding against zeroing, this time relying exclusively on Article 9.3 AD Agreement. In its report on US–Zeroing (EC), the Appellate Body avoided having to explain some of the problems of its earlier reliance on Article 2.4.2 AD Agreement by applying judicial economy in respect of the Article 2.4 claims of the EC.Footnote 15
Some of the challenges the Appellate Body encountered were clearly of its own making, not only because of the textual nature of its analysis but also because of some of the statements that accompanied the analysis. On a number of occasions, the Appellate Body explicitly stated that its rulings should be seen as limited to the weighted-average-to-weighted-average methodology before it; this in spite of the fact that the terms it was relying on and the reasoning used to support its findings against the use of zeroing clearly also applied to the second (transaction-to-transaction) methodology. This, as well, fueled ‘hopes’ in the pro-zeroing camp, and the Panel in US–Softwood Lumber V, Article 21.5, openly relied on such statements in support of its view that the Appellate Body did not wish to outlaw zeroing, irrespective of the methodology used, as had been argued by Canada.Footnote 16
The Appellate Body prides itself in restricting its findings to the case before it in an attempt to avoid accusations of it acting like a lawmaking body. The result however, in this string of cases, as on other occasions, is an absence of certainty and predictability. Continuous litigation over a rather technical and straightforward issue has wasted a lot of the WTO's resources as a consequence.
This criticism of the Appellate Body's approach in this matter should not be read to imply that the Panels, which found in favor of zeroing, were right. They were not, for the reasons briefly discussed earlier. While the Panels may have been correct to argue that a number of their technical arguments were not, or not convincingly, addressed by the Appellate Body, one may equally wonder why certain Panels opted to openly challenge the authority of the Appellate Body in this context. What explains the fact that these Panels were not afraid of the potential systemic implications of rejecting the unwritten rule of precedent that applies in WTO case law over this technical issue of zeroing? Why did Panels take on the Appellate Body, and the quasi entire WTO Membership for that matter, in the defense of a practice used really only by one WTO Member in antidumping investigations – a practice one has to accept is biased in favor of the domestic industry and is thus clearly unfair? These remain open questions.
The paper by Vermulst and Prusa discussing decisions by WTO Panels and the Appellate Body in respect of two comprehensive challenges of the practice of zeroing in antidumping proceedings explains, in a clear and thorough manner, how this zeroing methodology operates and the effect it has on the determination of dumping margins. Such a proper description of this practice and the effect it has on the determination and calculation of dumping margins is essential for understanding the legal discussion that has been raging at the WTO in respect of zeroing since 1999. Vermulst and Prusa thus provide an important and very useful contribution to the ongoing discussion about the use of zeroing in antidumping investigations, as Members continue to fight over this issue both in dispute settlement, and in the context of the Doha Round negotiations on a new WTO Antidumping Agreement (‘AD Agreement’).
The paper is very rich in terms of explaining the various ways in which zeroing can affect the margins of dumping, and the authors' clear and readable discussion of this technical question of antidumping law deserves a lot of praise. It is not my intention in this Commentary to repeat the excellent analysis of Vermulst and Prusa.Footnote 1 I will rather address a couple of the legal arguments and conclusions of Vermulst and Prusa that merit some additional comments. In addition, I will raise a more systemic question relating to the zeroing case law and examine briefly who is to be held responsible for the fact that this technical antidumping question is still being litigated now, almost ten years after it was first raised in WTO dispute settlement.
1. The law and economics of zeroing – worlds apart?
1.1. Zeroing and fairness
The economic analysis of the practice of zeroing in Vermulst and Prusa is clear: zeroing inflates the margins of dumping and allows an investigating authority to find dumping where no dumping would have been found and no duties would have been imposed otherwise. The conclusions to be drawn, after reading the explanation offered and the examples provided, is undoubtedly that zeroing is unfair and biased as it makes it more likely that dumping will be found to exist. Moreover, as clearly stated by Vermulst and Prusa, it is never the case that zeroing will lead to a lower dumping margin. In other words, zeroing is biased in favor of the petitioner industry and never benefits exporters. It is thus lacking in evenhandedness, and therefore, unfair.
The clear demonstration by Vermulst and Prusa of such lack of fairness contrasts with the legal analysis in which, somewhat surprisingly, they conclude that fairness is a concept that is ‘inherently vague’, suggesting that no finding of violation can be based on such a concept. Vermulst and Prusa state that ‘What is fair in the eyes of an exporter is not necessarily the same as what is fair in the eyes of a domestic producer.’ In so doing, Vermulst and Prusa are echoing the analysis of the Panels in the cases discussed, which, for that same reason, refused to consider that zeroing was unfair or violated the general requirement of Article 2.4 of making a fair comparison between normal value and export price.
In my view, the authors' own well-argued explanation of how the practice of zeroing inflates the margins of dumping and never works to the benefit of exporters is the best evidence of the fact that zeroing is inherently unfair and thus inconsistent with the disciplines of the Antidumping Agreement. The authors' correct economic analysis shows that zeroing is unfair. It seems difficult to accept that a legal analysis would lead to a different conclusion. Vermulst and Prusa seem to suggest that the concept of fairness is too vague and undefined to allow a legal finding to be based on such a concept. However, it is clear that WTO case law in respect of trade-remedy investigations is almost entirely based on similarly ‘vague’ concepts as Panels examine whether a ‘reasonable’ explanation has been provided as to how the facts support the determination made, whether a ‘determination’ was made in the case of reviews, or whether an examination of the injury factors was ‘objective’. In the case of zeroing, the question is not, as suggested by Vermulst and Prusa, whether exporters or the domestic producers consider that the practice of zeroing is unfair or not; but whether the Panel reaches this conclusion. ‘Fairness’ is clearly not a concept that is more vague than ‘reasonableness’ or ‘objectiveness’, which are necessarily employed by Panels all the time.
The main reason why both the Panels and the authors seem to be so keen to avoid the obvious legal conclusion that zeroing is unfair relates to the fact that it arguably would imply a violation of the requirement of Article 2.4 of making a fair comparison between normal value and export price. As the informed reader will know after having read the paper by Vermulst and Prusa, the discussion relating to zeroing started under a subparagraph of this provision, Article 2.4.2 AD Agreement, in which the three methodologies for comparing normal value and export price are set forth. All of the textual arguments made thus far that zeroing is not prohibited by the text of this provision would be rendered moot if, indeed, zeroing is considered as unfair under the generally applicable ‘fair comparison’ requirement of Article 2.4 to which Article 2.4.2 expressly refers. In order to maintain the viability of some of the textual arguments, as developed by certain WTO Members such as the United States and as adopted by a number of Panels under Article 2.4.2, one has to reach the bizarre conclusion that at least from a legal point of view, zeroing is not as ‘inherently unfair’ as the numbers tend to show. However, it seems more reasonable to conclude that what this internal contradiction reveals is that the textual arguments under Article 2.4.2 AD Agreement are simply not that strong. These textual arguments are contradicted by the commonsense requirement that an antidumping investigation be ‘unbiased and objective’ as required by Article 17.6(i) AD Agreement.
This is not to say that I would agree with the Appellate Body that zeroing is a violation of the ‘fair comparison’ requirement of Article 2.4 AD Agreement. Indeed, perhaps the more important comment to make in respect of the Appellate Body's finding in respect of the inconsistency of zeroing with the fair-comparison requirements of Article 2.4 AD Agreement is that this provision only requires an authority to make a fair comparison between normal value and export price but does not set forth a general fairness requirement. Arguably, zeroing has nothing to do with ensuring an apples-to-apples comparison, which is what is actually required by Article 2.4, as the problem is not related to the comparison of normal value with export price, but rather relates to the way the authority deals with the results of such comparisons as it zeroes the negative margins. In my view, Article 2.4 is merely an expression of the general underlying requirement that any antidumping investigation be objective and unbiased and that the laws and regulations relating to the conduct of antidumping investigations are equally conducive to such investigations. Zeroing is clearly at odds with this general requirement, which finds its expression, inter alia, in Article 17.6(i) of the AD Agreement. In sum, whether in law or in economics, zeroing is unfair and therefore inconsistent with the AD Agreement.
1.2. Zeroing and mathematical equivalence
The authors' legal analysis reaches a conclusion different from that reached under their economic analysis in respect of another important aspect of the zeroing debate as well. In their legal analysis, the authors seem to side with the views expressed by some Panels that zeroing cannot be considered as unfair and outlawed in a general manner as this would render inutile the third, exceptional method for dealing with targeted dumping, as set forth in Article 2.4.2 of the AD Agreement. This is the ‘mathematical equivalence’ argument, which posits that without zeroing, the third method (weighted-average normal value to transaction-specific export price) would lead to results that are mathematically equivalent to those obtained under the first method (a weighted-average normal value to weighted-average export-price comparison). Over time, this equivalence argument has become the prime argument to argue that zeroing must be permitted in certain circumstances, and cannot therefore be considered as generally ‘unfair’. The Panels on US–Zeroing (EC) and US–Zeroing (Japan) as well as US–Softwood Lumber V, Article 21.5 argued that the third methodology necessarily permits the use of zeroing, as only through zeroing nondumped export transactions would the result of this third methodology differ from the results reached by applying the first (weighted-average-to-weighted-average) methodology. If this is so, the Panels argued, then the Appellate Body's findings, even though based on terms that could relate to all three methodologies, must be limited to the first methodology, because extending the prohibition of zeroing to the second and third methodology would deprive of all meaning the third methodology. Effective treaty interpretation would argue against such a conclusion, and it cannot be assumed that the Appellate Body wanted to reduce to a nullity a provision of the Agreement. Moreover, according to the Panels, if zeroing is permitted under the third methodology, it cannot be considered to be unfair per se, or inconsistent with the general fair-comparison requirement of Article 2.4 AD Agreement.
The Appellate Body first rejected this mathematical-equivalence argument in its report on US–Softwood Lumber V, Article 21.5.Footnote 2 Interestingly, Vermulst and Prusa equally show that the basic premise of this argument is in error as it is simply not so that the third method becomes meaningless without zeroing. As stated in their paper, the exceptional method can produce different dumping amounts than the two ordinary methods, even without invoking zeroing. Yet, in the legal analysis, the authors keep repeating the argument that the Appellate Body has rendered the entire exceptional method inutile. It is clear that there may be circumstances where the law and the economics lead to different results. However, it seems that in a case like this one, where the legal argument is based on a quasi-economic justification of ‘mathematical equivalence’, the economic analysis that demonstrates the fallacy of this mathematical-equivalence argument must be given due weight. It seems difficult to continue to make this argument in law if the mathematics relied on to support this argument prove to be wrong.
2. Zeroing – merely a matter of transparency?
Another issue dealt with in the cases discussed by Vermulst and Prusa concerned the question whether the ‘zeroing methodology’ as consistently used by the United States can be challenged ‘as such’, independent from its actual application in a particular case. The discussion of whether such a methodology is a ‘measure’ and whether this measure is the line in the computer program used to calculate margins of dumping or rather the methodology in a more general manner is really much ado about nothing. The Appellate Body rightly considered that a methodology which is consistently used is challengeable and does not need to take the form of a law or regulation in order for it to be successfully challenged before the WTO. Vermulst and Prusa agree with the Appellate Body. However, they express the view that the US lost the claims against the zeroing methodology ‘as such’ because the US was ‘again the victim of its own transparency’. In this respect, they refer to another set of cases relating to a number of challenges against the US practice in sunset reviews, as set forth in the US Sunset Policy Bulletin (‘SPB’). In the paper by Vermulst and Prusa, it is suggested that transparency was also a key reason why the US lost several Appellate Body challenges to its sunset-review practice. The sunset case related to a consistent application of a particular methodology for determining a likelihood of recurrence or continuation of dumping in sunset reviews that would justify the extension of the antidumping measure for another five-year period.
It is interesting that the authors raise the Sunset Policy Bulletin cases in this context. Actually, it seems worth mentioning that there is an important difference in approach between the zeroing cases and these sunset cases. In the sunset cases, the US was actually able to rebut the challenges against its Sunset Policy Bulletin, which was never found to be WTO-inconsistent by the Appellate Body. The main reason why the Appellate Body refuses to find against the US in respect of the Sunset Policy Bulletin is because this policy document, with no binding force, contains language that suggests that the authority is not required to act in a manner which would be inconsistent with the requirement of the Antidumping Agreement, i.e. to base a sunset review on facts and not presumptions. The SPB provides for three scenarios that, if demonstrated to be present, are deemed to show that dumping is likely to continue or recur in the future.
The Appellate Body considered that such scenarios may be a useful tool, but their application should be case specific, and they should never be considered as determinative. However, even though the complainants were able to show that the US examines the likelihood of dumping in sunset reviews in 100% of the cases on the basis of the three scenarios of the Sunset Policy Bulletin, and does not examine any other factual evidence, and despite the absence of even one counterexample by the US, the Appellate Body considered that the Sunset Policy Bulletin was not WTO-inconsistent ‘as such’. The Appellate Body found that it had not been demonstrated that the authority considered that it was bound to follow the scenarios of the SPB or that it was precluded from examining other evidence. The 100% consistent application of the Bulletin was rejected by the Appellate Body in United States – Sunset Reviews of Anti-Dumping Measures on Oil Country Tubular Goods from Argentina as ‘mere statistical evidence’.Footnote 3
The difference in approach by the Appellate Body in respect of the zeroing methodology for which no written ‘policy document’ of any sort exists is striking. In respect of zeroing, the 100% consistent application of the methodology was considered determinative to establish the existence of a measure which is, ‘as such’, inconsistent with the WTO Agreements. In other words, it seems that if a WTO Member wants to avoid a finding of inconsistency of its practice, it is better to have some sort of policy document to refer to, as the effect of the mandatory/discretionary distinction is such that any potential for WTO-consistent behavior will allow a country to argue its way out of a 100% inconsistent application of the challenged norms.
Returning to the zeroing cases, it seems too easy to conclude that the US lost these cases because it was the victim of its own transparency. To argue that the US was the victim of its own transparency seems to suggest that transparency is a mitigating circumstance; it suggests that many WTO Members zero, but that the US simply does so openly. This is not correct. The general opposition in the course of the Doha Negotiations to proposals by the US to introduce a provision that expressly sanctions the use of zeroing is proof of the fact that zeroing is very much a US practice. The US was found to have violated the WTO Agreements not because it was too transparent, but because the zeroing methodology is WTO inconsistent. The transparent application of this methodology may well have made it easier to demonstrate the violation, but that is simply a matter of evidence.
3. The zeroing saga – a systemic failure?
Although it is now clearly settled that zeroing is not permitted under the AD Agreement, the road to this conclusion has been particularly thorny. In total, ten casesFootnote 4 and 16 reports so far,Footnote 5 have dealt with the issue of zeroing. The zeroing case law is made up of diverging Panel reports, two dissenting opinions,Footnote 6 a Panel overturning itself,Footnote 7 the Appellate Body changing its reasoning to deal with the issue on several occasions, and includes a number of Panels openly disagreeing with the Appellate Body and refusing to follow its reasoning.Footnote 8 The zeroing saga is still continuingFootnote 9 in spite of what is now a clear and comprehensive set of rulings outlawing zeroing. This begs the question who is to blame for this seemingly endless string of cases dealing with something that is, after all, a very technical aspect of antidumping investigations. While Vermulst and Prusa do not address this question, I would like to offer a couple of reflections in this respect.
In a bizarre way, the cases discussed by Vermulst and Prusa and the other cases relating to the practice of ‘zeroing’ reveal some of the problems in terms of securing predictability and security that affect the Appellate Body's jurisprudence. The zeroing case law started with a Panel adopting a position as close as it gets to a rule of binding precedent (stare decisis) and (for now) ended with Panels for the first time openly refusing to follow the Appellate Body's reasoning and rejecting the Appellate Body's interpretation.
The search for the correct legal basis on which to analyze the practice of zeroing is what characterizes the zeroing case law. It is clear that the Antidumping Agreement does not contain an express prohibition or permission of zeroing. Actually, the term is not mentioned in the Antidumping Agreement. This does not necessarily mean that zeroing is permitted under the Agreement, as became clear from the case law. Still, the problem the Panels and the Appellate Body obviously struggled with was finding the right textual hook to hang the analysis on.
The textual basis for much of the Panels' and the Appellate Body's analysis was Article 2.4.2 AD Agreement dealing with the methods for comparing normal value and export price in order to establish a margin of dumping, and the general requirement to conduct a fair comparison between normal value and export price of Article 2.4 AD Agreement.Footnote 10
The basic problem and the reason for the fact that the zeroing discussion dragged on for so long was the fact that, for too long, the Appellate Body failed to call the thing by its name. It is clear that the practice of zeroing is at odds with the requirement to conduct an objective and unbiased investigation into the existence of dumping.Footnote 11 The requirement of evenhandedness is crucial in conducting an objective examination, and, as the Appellate Body has made clear on numerous occasions, the use of a methodology which makes it more likely that the investigating authority will determine that there is dumping, as is the case with zeroing, is at odds with this obligation.Footnote 12
In an antidumping investigation, there will always be many transactions that will need to be compared, sometimes for different models. So, under any method, the investigating authority will need to calculate an average in the end, whether it is the average of the results of all comparisons of export transactions with domestic transactions, or the average of the results of weighted-average-to-weighted-average comparisons for different models. Averaging precisely implies that positive results will be offsetting negative results. By zeroing, the investigating authority tinkers with the results by disallowing the positive results to fully offset the negative results, thus making it more likely, if not almost inevitable, that a dumping margin will be found. Given this basic problem with zeroing, it is clear that in the absence of an explicit authorization in the text of the AD Agreement that would allow for this type of distorted examination, zeroing has to be considered as prohibited under the AD Agreement. The argument that an explicit prohibition was discussed, but no agreement was reached during the Uruguay Round for the inclusion of such a prohibition, does not convince.Footnote 13 Actually, the fact that zeroing was discussed and no explicit authorization of this obviously biased practice was included in the AD Agreement, only confirms its illegitimate status.
The problem was that the Appellate Body, prior to its report on US–Softwood Lumber V, Article 21.5, had, with one notable exception, shied away from making this basic point that would have ended all further discussion on this matter. Indeed, on one occasion, in its report on US–Corrosion-Resistant Steel Sunset Review, a dispute that only indirectly related to the question of zeroing, the Appellate Body stated in a very straightforward manner that zeroing is inherently unfair and therefore inconsistent with the AD Agreement, irrespective of whether it is applied in original investigations or in reviews:
In EC–Bed Linen, we upheld the finding of the panel that the European Communities acted inconsistently with Article 2.4.2 of the Anti-Dumping Agreement by using a ‘zeroing’ methodology in the anti-dumping investigation at issue in that case. We held that the European Communities' use of this methodology ‘inflated the result from the calculation of the margin of dumping’. We also emphasized that a comparison such as that undertaken by the European Communities in that case is not a ‘fair comparison’ between export price and normal value as required by Articles 2.4 and 2.4.2.
When investigating authorities use a zeroing methodology such as that examined in EC–Bed Linen to calculate a dumping margin, whether in an original investigation or otherwise, that methodology will tend to inflate the margins calculated. Apart from inflating the margins, such a methodology could, in some instances, turn a negative margin of dumping into a positive margin of dumping. As the Panel itself recognized in the present dispute, ‘zeroing … may lead to an affirmative determination that dumping exists where no dumping would have been established in the absence of zeroing’. Thus, the inherent bias in a zeroing methodology of this kind may distort not only the magnitude of a dumping margin, but also a finding of the very existence of dumping.Footnote 14 (footnotes omitted) (emphasis added)
In all of the cases more specifically dealing with zeroing that followed this case, the Appellate Body was looking for a textual hook to prohibit zeroing and failed to outlaw zeroing for what it was, biased and unfair. The textual arguments of the Appellate Body proved to be problematic and became a source of as much discussion as, if not more than, the actual text of the AD Agreement. A number of Panels, as well as some dissenting panelists, all pointed to a couple of technical and textual shortcomings of the Appellate Body's arguments.
The Appellate Body had to move away from the ‘all comparable transactions’ language relied on in EC–Bed Linen to avoid being trapped in a prohibition that related exclusively to the first methodology of Article 2.4.2 AD Agreement. In US–Softwood Lumber V, it thus relied on the term ‘margin of dumping’ as it appeared at the beginning of the opening sentence of Article 2.4.2 AD Agreement. However, its reasoning was still very much linked to Article 2.4.2 AD Agreement. The Panel in US–Zeroing (EC) pointed to the fact that Article 2.4.2 limits its application to the ‘investigation phase’, and on that basis followed the US in its argument that zeroing was not prohibited in the context of administrative reviews. Rather than dealing with this textual, albeit unconvincing argument by the Panel, the Appellate Body again changed the basis for its finding against zeroing, this time relying exclusively on Article 9.3 AD Agreement. In its report on US–Zeroing (EC), the Appellate Body avoided having to explain some of the problems of its earlier reliance on Article 2.4.2 AD Agreement by applying judicial economy in respect of the Article 2.4 claims of the EC.Footnote 15
Some of the challenges the Appellate Body encountered were clearly of its own making, not only because of the textual nature of its analysis but also because of some of the statements that accompanied the analysis. On a number of occasions, the Appellate Body explicitly stated that its rulings should be seen as limited to the weighted-average-to-weighted-average methodology before it; this in spite of the fact that the terms it was relying on and the reasoning used to support its findings against the use of zeroing clearly also applied to the second (transaction-to-transaction) methodology. This, as well, fueled ‘hopes’ in the pro-zeroing camp, and the Panel in US–Softwood Lumber V, Article 21.5, openly relied on such statements in support of its view that the Appellate Body did not wish to outlaw zeroing, irrespective of the methodology used, as had been argued by Canada.Footnote 16
The Appellate Body prides itself in restricting its findings to the case before it in an attempt to avoid accusations of it acting like a lawmaking body. The result however, in this string of cases, as on other occasions, is an absence of certainty and predictability. Continuous litigation over a rather technical and straightforward issue has wasted a lot of the WTO's resources as a consequence.
This criticism of the Appellate Body's approach in this matter should not be read to imply that the Panels, which found in favor of zeroing, were right. They were not, for the reasons briefly discussed earlier. While the Panels may have been correct to argue that a number of their technical arguments were not, or not convincingly, addressed by the Appellate Body, one may equally wonder why certain Panels opted to openly challenge the authority of the Appellate Body in this context. What explains the fact that these Panels were not afraid of the potential systemic implications of rejecting the unwritten rule of precedent that applies in WTO case law over this technical issue of zeroing? Why did Panels take on the Appellate Body, and the quasi entire WTO Membership for that matter, in the defense of a practice used really only by one WTO Member in antidumping investigations – a practice one has to accept is biased in favor of the domestic industry and is thus clearly unfair? These remain open questions.