1. Introduction
On 9 January 2007, the Appellate Body (AB) issued its report in United States – Measures Relating to Zeroing and Sunset Reviews (Japan).Footnote 1 This was the fifth AB report in which some aspect of zeroing was adjudicated,Footnote 2 making it the single most litigated subject in the history of the WTO.
This paper covers both this report and the AB report in US–Zeroing (EC).Footnote 3 Both cases involved the US Department of Commerce's practice of zeroing and raised very similar issues that justify this joint review. It is organized as follows. In Sections 2, 3, and 4, we will first discuss the facts of the cases and the claims of the parties. This will be followed by Section 5, which will provide an overview of the Panel and AB findings in these cases. Thereafter, in Section 6, we will provide our detailed legal and economic analysis on zeroing. Lastly, by way of conclusion, certain thoughts on the future of zeroing will be presented.
The scope of the claims in both cases was considerably broader than in the previous WTO disputes involving zeroing. We conclude that, while the Appellate Body was arguably correct in prohibiting the use of zeroing under the main methods of Article 2.4.2 of the Antidumping AgreementFootnote 4 [ADA] as well as in various reviews, it overreached in considering zeroing to be in violation of the fair-comparison requirement of Article 2.4 AD Agreement and as inconsistent with Article 2.4.2, exceptional method. Furthermore, while the AB found zeroing in reviews violated Article 2.4.2 AD Agreement, we believe it would have been preferable for the AB to have limited its findings of inconsistency to Article 9.3 AD Agreement.
2. The facts
2.1 US–Zeroing (EC)
In the US–Zeroing (EC) dispute, which originated in June 2003, the EC essentially challenged two types of zeroing. First, the EC contested the use of model zeroing by the United States Department of Commerce (USDOC) while calculating the dumping margin on the basis of the weighted-average-to-weighted-average comparison method in original investigations. Secondly, the EC objected to the use of simple zeroing by the USDOC while calculating the antidumping margins on the basis of a weighted-average-to-transaction comparison method for the assessment of an importer's final liability for paying antidumping duties and future cash-deposit rates (i.e., administrative reviews).Footnote 5
The EC used the term model zeroing to refer to the exclusion from the numerator of the weighted-average dumping margin of any amounts by which average export prices within individual averaging groups based upon physical characteristics exceeded the average normal values. As regards simple zeroing, the EC meant the practice by which the USDOC while aggregating the results of comparisons between the normal value and export prices made on an average-to-transaction basis excludes from the numerator of the weighted-average dumping margin any amounts by which individual export prices exceed the weighted-average normal values.
Accordingly, the EC challenged certain US legal instruments, procedures, methodologies, and practices in relation to these different types of zeroing ‘as such’ and ‘as applied’ as being contrary to the provisions of the WTO Antidumping Agreement (AD Agreement),Footnote 6 GATT 1994, and the WTO Agreement. In particular, it contested: (1) model zeroing in 15 ‘as applied’ cases in original antidumping investigations; (2) simple zeroing in 16 ‘as applied’ cases in periodic antidumping reviews (administrative reviews); (3) sections 731, 751(a)(2)(A)(i) and (ii), 771(35)(A) and (B), and 777A(d) of the Tariff Act of 1930, as amended, (Tariff Act) ‘as such’; (4) section 351.414(c)(2) of the USDOC Regulations, ‘as such’; (5) certain provisions of the 1997 edition of the Import Administration Antidumping Manual, ‘as such’; (6) the Standard AD Margin Program which includes standard zeroing procedures, ‘as such’; and (7) the US practice or methodology of zeroing, ‘as such’.Footnote 7
2.2 US –Zeroing (Japan)
Similar to the EC's claims in US–Zeroing (EC), in November 2004, Japan also challenged the US's use of model zeroing and simple zeroing in antidumping proceedings. It alleged that the US applied a methodology for the calculation of an overall dumping margin that excluded the amounts by which export prices for certain transactions exceeded the normal value and referred to this as the zeroing procedures and the Standard Zeroing LineFootnote 8 in the USDOC's dumping-calculation software program.
Japan defined model zeroing in the same manner as the EC, and it asserted that the USDOC routinely applies model zeroing in original investigations but had also resorted to simple zeroing in an original investigation while using the transaction-to-transaction comparison method. Moreover, it alleged that the USDOC also routinely uses simple zeroing in average-to-transaction comparisons in administrative reviewsFootnote 9 and new-shipper reviews. It defined simple zeroing as the USDOC's general methodology of calculating the weighted-average dumping margin on the basis of an average-to-transaction comparison or transaction-to-transaction comparison between the export price and the normal value by excluding (while aggregating the results of these multiple comparisons) all amounts by which the export prices of individual transactions exceed the normal value. Lastly, Japan also argued that in changed-circumstances reviews and sunset reviews, the USDOC generally relies on dumping margins previously calculated in original investigations using model zeroing or on dumping margins calculated in administrative reviews using simple zeroing.
In sum, Japan challenged ‘as such’ these zeroing procedures, as a measure pursuant to the AD Agreement and the Dispute Settlement Understanding (DSU) and also their application with respect to Japanese products in one original investigation, 11 administrative reviews, and two sunset reviews.
3. Claims
3.1 EC's claims in US–Zeroing (EC)
3.1.1 Claims with respect to original investigation
3.1.1.1 The ‘as applied’ claims regarding the inconsistency of model zeroing with Article 2.4.2 AD Agreement in the context of 15 original investigations
The first claim of the EC targeted the use of ‘model zeroing’ by the US in 15 original investigations as being inconsistent with Article 2.4.2 AD Agreement. The EC considered that excluding the results of the comparisons in which the weighted-average export prices exceeded the weighted-average normal values from the calculation of the overall dumping margin was contrary to Article 2.4.2 AD Agreement. To support its assertion, the EC firstly referred to the Panel and Appellate Body reports in EC–Bed Linen Footnote 10 and US–Softwood Lumber V Footnote 11 and argued that the words ‘margins of dumping’ used in Article 2.4.2 AD Agreement apply to the product under investigation as a whole and not to models, types, or categories of such products because Article 2.1 of the AD Agreement defines ‘dumping’ in relation to a product. Therefore, the intermediate margins calculated by the USDOC for specific models do not constitute dumping margins as per Article 2.4.2 AD Agreement. Secondly, it underlined that the requirement of Article 2.4.2 AD Agreement is that of a simple comparison between the export price and the normal value and any difference between the two, whether positive or negative, constitutes a ‘margin’ within the meaning of the said provision.
Therefore, according to the EC, since the USDOC had defined the product subject to the investigation, it should have calculated a dumping margin for the product as a whole by comparing the weighted-average export price of all transactions within individual averaging groups including export prices above normal value and accordingly should have incorporated the negative margins of particular averaging groups as well.
Against this claim, the EC also raised certain consequential claims as regards the inconsistency of the US measures with Articles 1, 3.1, 3.2, 3.5, 5.8, 9.3, and 18.4 AD Agreement, Articles VI:1 and VI:2 GATT, and Article XVI:4 WTO Agreement.Footnote 12
3.1.1.2 The ‘as such’ claims with respect to the standard zeroing procedures and certain provisions of the Tariff Act
3.1.1.2.1 The ‘as such’ claims with respect to the Tariff Act
The EC asserted that the criterion as regards the ‘as such’ claims is whether the measure in question is in conformity with the AD Agreement. Therefore, if the domestic law requires the application of measures that are contrary to the AD Agreement, the cause of the inconsistency partly lies in these domestic laws. Accordingly, the EC challenged as measures sections 771(35)(A) and (B) Tariff Act (which provide for a definition of ‘dumping margin’ and ‘weighted average dumping margin’ respectively); section 731 Tariff Act (which provides rules regarding the imposition of antidumping duties); and section 777A(d) Tariff Act (which provides the determination of ‘Less Than Fair Value’) on their consistency ‘as such’ with Articles 2.4, 2.4.2, 5.8, 9.3, 1, and 18.4 AD Agreement, Articles VI:1 and VI:2 GATT, and Article XVI:4 WTO Agreement, claiming that the USDOC had repeatedly asserted that zeroing is required by these provisions.
The EC supported its contention by emphasizing that in considering the WTO-consistency of discretionary legislation, its effects must be taken into account and a determination must be made to delineate if it is the law or more generally the document containing the general rule of prospective application that is the cause of the inconsistency. Therefore, discretionary legislation like the Tariff Act, which is multi-interpretable and is consistently interpreted and applied and allows WTO-inconsistent behavior, could be found to be inconsistent ‘as such’.Footnote 13
3.1.1.2.2 The ‘as such’ claims with respect to the standard zeroing procedures
Considering that any omission or act of a WTO member including acts prescribing rules or norms that are intended to have general and prospective application can be challenged as a measure in dispute settlement, the EC argued that measures that include the standard zeroing procedures or the specific lines of the computer code contained in one of the computer programs (i.e., AD Margin Program, which contains the USDOC's dumping-margin-calculation methodology by which negative and positive margins are separated and only the dumping amounts for sales with positive margins are aggregated) and the consistent US practice of zeroing violate Articles 2.4 and 2.4.2, 5.8, 9.3, 1, 18.4 AD Agreement, Articles VI:1 and VI:2 GATT, and Article XVI:4 WTO Agreement ‘as such’. The EC considered it relevant to evaluate the WTO-consistency of the ‘as such’ claims against these measures irrespective of their mandatory or discretionary character. It claimed that the standard zeroing procedures and the Antidumping Manual containing these procedures are administrative procedures, within the meaning of Article 18.4 AD Agreement, and they have some legal effect and are in practice treated by the USDOC as binding. More specifically, it alleged that the standard zeroing procedures have a legal character since the USDOC cannot depart from them without a reasoned justification, and the computer program, i.e. the AD Margin Program, is published by the USDOC, and it establishes rules generally referred to as standards and source of guidance by USDOC officials. Thus, the EC considered these to be WTO-inconsistent ‘as such’ because they provide for the application of a standard different from that set out in Article 2.4.2 AD Agreement and stated that it was evident that the practice of zeroing had been incorporated by the US investigating authority itself in the AD Margin program and the Antidumping Manual.
3.1.2 Claims with respect to administrative reviews
The EC claimed the violation of WTO obligations by the US in administrative-review proceedings pertaining to the review of the antidumping duty orders in which the USDOC determines the ‘percentage weighted average dumping margins’ and ‘cash deposit rates’ for individual exporters/producers and assessment rates for individual importers.Footnote 14
3.1.2.1 Claims with respect to 16 administrative-review proceedings
3.1.2.1.1 Claims regarding the inconsistency of simple zeroing with Articles 2.4.2 and 2.4 AD Agreement
First, the EC argued that by using the asymmetrical average-to-transaction comparison method and not using the two symmetrical methods when the conditions for using the former were not fulfilled, the USDOC acted in contravention of Article 2.4.2 AD Agreement in 16 administrative-review proceedings. Secondly, it contested the use of simple zeroing in the aggregation of the comparison results in order to calculate an overall dumping margin as violating Article 2.4.2 AD Agreement.
As regards the first claim, the EC asserted that Article 2.4.2 AD Agreement applies in the context of investigations as well as reviews because it applies to any proceeding under the AD Agreement in which antidumping margins are established. Addressing the meaning of the word ‘investigation’, it considered that investigation is not limited to the explicit words of a provision as in Article 5 AD Agreement referring to the determination, existence, degree, and effect of alleged dumping; rather, Articles 9.3.1, 9.5, 11.2, and 11.3 AD Agreement also require authorities to conduct investigations. The EC provided several interpretations of the words ‘during the investigation phase’, as used in Article 2.4.2 AD Agreement, to demonstrate that the obligations contained therein are not limited to original investigations and also claimed that the word ‘phase’ as mentioned in the said provision does not support the view that ‘investigation’ defined in Article 5.1 AD Agreement is for the purpose of the entire AD Agreement. The EC supported its latter claim on the basis of the Appellate Body and Panel reports, the negotiating history of the AD Agreement, and the international norms of treaty interpretation. It considered that the nature of the activity of the investigating authority and not the scope of the inquiry determines if it is an investigation and argued that an assessment proceeding under Article 9.3.1 AD Agreement entailing a systematic examination of the degree of dumping is an investigation. Therefore, it contested that the nonapplication of Article 2.4.2 AD Agreement would lead to unequal treatment between prospective and retrospective duty systems. The reason being, that in the prospective system the dumping margin is established on the basis of the original investigation in which zeroing is prohibited, whereas in the retrospective system the dumping margin established in the initial investigation prohibiting zeroing will be eclipsed by duty determinations using zeroing in administrative reviews. This, in turn, will lead to higher duties in the latter system.
Concerning its claims under Article 2.4 AD Agreement, the EC argued that besides using the asymmetrical average-to-transaction method, the US excluded from the numerator of the dumping margin all amounts by which prices of individual export transactions exceeded the normal values and this violated the fair-comparison obligation enshrined in the said provision. The EC claimed that the qualification of the comparison as ‘fair’ denotes a comparison that treats export and domestic sales symmetrically, and it necessarily precludes zeroing, which involves an artificial reduction of the prices and an inflation of the dumping margins. The EC asserted that zeroing is inconsistent with the first, third, and fifth sentences of Article 2.4 AD Agreement, because it permits an allowance that leads to a reduction of the export price for a difference other than that affecting price comparability and is thus not a ‘due allowance’. The EC elaborated its assertion that the first sentence of Article 2.4.2 AD Agreement stipulates that, in normal circumstances, fair comparison as per Article 2.4 AD Agreement implies symmetrical treatment, while the second sentence of Article 2.4.2 AD Agreement stipulates that in case of targeted dumping, the symmetrical treatment obligation can be departed from.
According to the EC, zeroing is inconsistent with Article 2.4 AD Agreement when the transaction-to-transaction method is used to calculate the dumping margin, but zeroing is not unfair when the average-to-transaction method is used, provided that the conditions for the use of the latter are fulfilled. Moreover, it claimed that Article 9.2 AD Agreement does not support that asymmetrical comparison methods be permitted in the duty-assessment phase.
3.1.2.1.2 The ‘as applied’ claims under other provisions of the AD Agreement; Articles VI:1, VI:2 GATT; and Article XVI:4 WTO Agreement in relation to certain administrative reviews
The EC also contested the actual application of simple zeroing by the USDOC in 16 administrative reviews as violating Articles 1, 9.3, 11.1, 11.2, 18.4 AD Agreement, Articles VI:1, VI:2 GATT, and Article XVI:4 WTO Agreement.
3.1.2.1.3 The ‘as such’ claims against the standard zeroing procedures, the Tariff Act, and USDOC Regulations in relation to administrative reviews
The EC also raised certain consequential claims and contested that the standard zeroing procedures; the US's practice of zeroing, sections 771(35)(A) and (B), 731, 777A(d), and 751(a)(2)(i) and (ii) Tariff Act; and section 351.414(c)(2) USDOC Regulations in the context of administrative reviews are inconsistent ‘as such’ with Articles 1, 2.4, 2.4.2, 9.3, 11.1, 11.2, 18.4 AD Agreement; Articles VI.1, VI:2 GATT; and Article XVI:4 WTO Agreement.
3.1.2.1.4 The ‘as such’ claims against the standard zeroing procedures in relation to new-shipper reviews, changed-circumstances reviews, and sunset reviews
The EC alleged that the standard zeroing procedures of the USDOC; sections 771(35)(A) and (B), 731, 777A(d), and 751(a)(2)(i) and (ii) Tariff Act; and section 351.414(c)(2) USDOC Regulations, in the context of new-shipper reviews, changed-circumstances reviews, and sunset reviews infringe ‘as such’ Articles 1, 2.4, 2.4.2, 9.3, 9.5, 11.1, 11.2, 11.3, 18.4 AD Agreement; Articles VI:1, VI:2 GATT; and Article XVI:4 WTO Agreement.
3.2 Japan's claims in US–Zeroing (Japan)
3.2.1 Claims that zeroing procedures are measures subject to dispute settlement under the DSU and the AD Agreement
Japan asserted that measures that can be challenged under the AD Agreement comprise any act or omission by a WTO Member, irrespective of their domestic status and binding/nonbinding character. Accordingly, it claimed that model and simple zeroing procedures are incorporated in a standard zeroing line or a line of a computer-programming code (AD Margin Program) and are specific measures within the meaning of Article 6.2 DSU. Additionally, it asserted that they are administrative procedures within the meaning of Article 18.4 AD Agreement, because computer instructions are covered by the ordinary meaning of the term ‘administrative procedures’. Japan asserted further that model and simple zeroing procedures constitute a predetermined, standardized method for conducting and managing USDOC's dumping-margin calculations, irrespective of the comparison method used in antidumping proceedings. It supported this contention by proffering different categories of evidence to demonstrate that zeroing procedures are a rule/norm of general and prospective application and have been consistently used by the US in every dumping-margin calculation in the past decade including the 26 cases at issue. Japan also referred to the Appellate Body's findings in US–Zeroing (EC), confirming model and simple zeroing to be measures challengeable ‘as such’.
3.2.2 Claims with respect to zeroing in original investigations
3.2.2.1 The ‘as such’ claims with respect to zeroing in original investigations
3.2.2.1.1 Claims with respect to the inconsistency ‘as such’ of model and simple zeroing with Articles 2.1, 2.4.2 AD Agreement and Articles VI:1, VI:2 GATT
Japan argued that it stems from the definition of ‘dumping’ and ‘margin of dumping’ in Article 2.1 AD Agreement and Article VI GATT that a dumping margin is to be established for a product as a whole. To support this argument it referred to the Appellate Body's analysis in US–Softwood Lumber V regarding the applicability of Article 2.1 AD Agreement to the whole AD Agreement (including Article 2.4.2) and asserted that Article 9.2 AD Agreement provides for the imposition of antidumping duties for the product as a whole. Therefore, by using model zeroing in average-to-average comparisons in original investigations, the USDOC disregards the negative results of comparisons and, as a result, all comparisons are not taken into account, and the dumping determination and dumping-margin calculation is not for the product as a whole. Japan also applied this reasoning to challenge simple zeroing used by the USDOC in original investigations. It claimed that the only exception to the fundamental rule of establishing a dumping margin for the product as a whole is under the average-to-transaction method provided in the second sentence of Article 2.4.2 AD Agreement, and prohibition of zeroing does not render that comparison method redundant.
Noting that the Appellate Body in US–Zeroing (EC) did not investigate whether zeroing was prohibited under the transaction-to-transaction comparison method in an original investigation, Japan said that the Appellate Body's reasoning dictated that conclusion. It confirmed that the requirement to determine ‘dumping’ and dumping margins for the product under investigation as a whole is applicable throughout the AD Agreement.
3.2.2.1.2 Claims with respect to the ‘as such’ inconsistency of model and simple zeroing with Article 2.4 AD Agreement
Japan claimed that both types of zeroing violated the fair-comparison obligation of Article 2.4 AD Agreement, which applies in the context of the entire Article 2 AD Agreement, including the calculation of the dumping margin, and requires the investigating authorities to identify the price difference between the normal value and the export price of a product in an unbiased manner. In support of this claim, Japan argued that the use of model and simple zeroing precludes an even-handed comparison of the export prices and normal value and leads to a finding of dumping even when there is no dumping. It added that the comparison is also distorted because the export prices are treated as less than what they actually are, and a dumping margin for the product as a whole is not determined. Lastly, Japan submitted that the prohibition of zeroing does not make the average-to-transaction comparison method redundant. It contested that without zeroing the average-to-transaction comparison would necessarily produce the same results as an average-to-average comparison, but different results will be generated if the average normal value is calculated on a different basis. In this regard, the second sentence of Article 2.4.2 AD Agreement addresses the comparison of an average normal value to only those export transactions that constitute the pattern of targeted dumping.
3.2.2.1.3 Claims with respect to the ‘as such’ inconsistency of model and simple zeroing with Articles 3.1 to 3.5, 5.8, 1, 18.4 AD Agreement and Article XVI:4 WTO Agreement
Japan claimed a violation of Articles 3.1 to 3.5 AD Agreement, due to the maintenance of zeroing procedures by the US in original investigations, because this leads to the systematic distortion of the dumping margin and consequentially the injury determinations are not objective (as they are not based on the volume and prices of dumped and nondumped imports, the rate of increase of dumped imports, and the magnitude of the margin of dumping). As regards Article 5.8 AD Agreement, Japan alleged that zeroing makes it impossible for the USDOC to determine adequately if there is sufficient evidence of dumping to justify the continuation of an antidumping investigation. With regard to model zeroing in original investigations, Japan claimed that the USDOC violates Articles 1 and 18.4 AD Agreement and Article XVI:4 WTO Agreement.
3.2.2.2 The ‘as applied’ claims with respect to the use of model zeroing in the proceedings concerning certain cut-to-length carbon-quality steel products
Japan challenged the application of model zeroing by the USDOC in the antidumping investigation of certain cut-to-length carbon-quality steel products from Japan as being inconsistent with Articles 2.1 and 2.4.2 AD Agreement and Articles VI:1, VI:2 GATT, because the US did not determine the existence of dumping nor did it calculate a dumping margin for the product as a whole. Secondly, Japan alleged that by using model zeroing in this investigation, the US had violated the fair-comparison obligation of Article 2.4 AD Agreement and also that the USDOC infringed Articles 3.1 and 3.5 AD Agreement, because model zeroing distorted the dumping determination and led to an injury determination that was not based on an objective examination of positive evidence.
3.2.3 Claims with regard to zeroing procedures in administrative reviews and new-shipper reviews
3.2.3.1 Claims with respect to simple zeroing ‘as such’ in administrative reviews and new-shipper reviews
3.2.3.1.1 Claims with respect to the inconsistency of simple zeroing ‘as such’ with Articles 2.1 and 2.4.2 AD Agreement and Articles VI:1, VI:2 GATT
Japan claimed that by applying simple zeroing in administrative reviews conducted pursuant to Article 9.3 AD Agreement and new-shipper reviews under Article 9.5 AD Agreement, the USDOC violates Articles 2.1 and 2.4.2 AD Agreement and Articles VI:1 and VI:2 GATT, because the determination of ‘dumping’ and the dumping-margin calculation is not for the product as a whole. It held that the dumping-margin calculation under Article 9.3 AD Agreement is subject to Article 2 AD Agreement including Article 2.4.2. Besides invoking other textual arguments,Footnote 15 Japan referred to US–Zeroing (EC) to support its assertion that simple zeroing in administrative reviews is WTO-inconsistent and also emphasized the uniform meaning of concepts like ‘dumping’, ‘margins of dumping’, and ‘product’ throughout the AD Agreement.
3.2.3.1.2 Claims with respect to the ‘as such’ inconsistency of simple zeroing with Articles 2.4, 9.1–9.3, 9.5, 1, 18.4 AD Agreement and Article XVI:4 WTO Agreement
Japan raised the same arguments with respect to the inconsistency of simple zeroing in administrative and new-shipper reviews with Article 2.4 AD Agreement as it did in the context of original proceedings. Further, referring to its previous claims with regard to Articles 2.1, 2.4, and 2.4.2 AD Agreement, Japan submitted that simple zeroing in administrative and new-shipper reviews is inconsistent with Articles 1, 9.1, 9.2, 9.3, 9.5, 18.4 AD Agreement and Article XVI:4 WTO Agreement.
Japan added that since the zeroing procedures inflate the dumping margin, the antidumping duty assessed and collected exceeds the margins that should have been calculated without zeroing.Footnote 16 Moreover, as zeroing infringes Article 2 AD Agreement, maintaining zeroing for determining dumping margins in administrative and new-shipper reviews violates Articles 9.1, 9.2, 9.3 AD Agreement and Article 9.5 AD Agreement respectively.
3.2.3.2 The ‘as applied’ claims with respect to the use of simple zeroing in administrative reviews
Japan argued that because of the use of simple zeroing in 11 administrative reviews, the USDOC infringed Articles 1, 2.1, 2.4.2, 2.4, and 9.1 to 9.3 AD Agreement and Articles VI:1, VI:2 GATT. It raised the same arguments as in the case of its ‘as such’ claims relating to administrative reviews and new-shipper reviews.
3.2.4 The ‘as such’ and ‘as applied’ claims with regard to zeroing procedures in changed-circumstances reviews and sunset reviews
Japan alleged that the US violates Articles 2 and 11 AD Agreement in changed-circumstance and sunset reviews ‘as such’ because in conducting these reviews the USDOC relies on dumping margins calculated in the original investigations or in administrative reviews using either model or simple zeroing. Referring to the Appellate Body's decision in US–Corrosion-Resistant Steel Sunset Review,Footnote 17 Japan claimed that in the above-mentioned reviews, if the USDOC relies on the previously calculated margins, such margins should be calculated consistently with the requirement of Articles 2.1, 2.4, and 2.4.2 AD Agreement to determine ‘dumping’ and dumping margin for the product as a whole. Therefore, Japan claimed that the US violates Articles 11.2 and 11.3 AD Agreement and consequentially Article 11.1 AD Agreement, which requires that ‘anti-dumping duties remain in force only as long as, and to the extent necessary, to counteract dumping’.Footnote 18
Furthermore, Japan also challenged the antidumping measures adopted in the two sunset reviews as being inconsistent with Articles 2 and 11 AD Agreement because the USDOC and USITC (United States International Trade Commission) relied on previously calculated dumping margins using the standard zeroing procedures.
4. US response
4.1 US response to the EC's claims in US–Zeroing (EC)
4.1.1 US response to the EC's claims with respect to zeroing in original investigations
4.1.1.1 Response to EC's claims regarding the consistency of model zeroing with Article 2.4.2 AD Agreement ‘as applied’ in 15 original investigations
The US counterargued that contrary to the EC's claims, Article 2.4.2 AD Agreement does not contain an obligation to calculate an overall dumping margin and cannot be read as imposing an obligation to offset negative dumping. It contended that, firstly, the term ‘margins of dumping’ in Article 2.4.2 AD Agreement does not apply to the product under investigation as a whole only. Basing its arguments on the view of the drafters of the AD Agreement and GATT, it asserted that dumping margin refers to both – ‘the results of particular comparisons between normal value and export price’ and to the ‘overall results of those comparisons’.Footnote 19 The US considered that in US–Softwood Lumber V, the Appellate Body had erred in finding that the AD Agreement requires the calculation of dumping margins in original investigations by taking into consideration those weighted-average comparisons where the export price is in excess of the normal value. Secondly, the US relied on the negotiating history of Article 2.4.2 AD Agreement to argue that this provision ‘restricts the application of the average-to-transaction comparison method in the investigation phase but does not address the offsetting of negative dumping’Footnote 20 and that, therefore, the EC's consideration of the concept of negative dumping margins lacks support from Article VI GATT or the AD Agreement.
Lastly, the US rejected the allegation that it excludes nondumped transactions from the calculation of an overall margin of dumping under the average-to-average comparison method because all the comparable export transactions are included in the averaging groups when taken together and also because the total figure by which the aggregate dumping amount is divided includes all the export transactions.
4.1.1.2 US response to the EC's ‘as such’ claims against zeroing with respect to the Tariff Act and the standard zeroing procedures
The US denied the ‘as such’ inconsistency of sections 771(35)(A) and (B) and 777A(d) Tariff Act and claimed that these sections do not prohibit the USDOC from offsetting the nondumped transactions and the EC could not prove such a prohibition. Secondly, it noted that the US Court of Appeals for the Federal Circuit had held twice that the Tariff Act did not oblige the use of zeroing. The US supported its argument further by asserting that a measure can be considered WTO-inconsistent only if it permits WTO-inconsistent action or prohibits WTO-consistent action and no Panel or Appellate Body has ever found a measure to be WTO-inconsistent if either of these conditions do not exist.
With respect to EC's claims against the standard zeroing procedures, the US contested that these are either not measures at all or not mandatory measures. Firstly, it claimed that the Antidumping Manual is not a legally binding instrument but just a guide for the USDOC officials, and it neither permits nor precludes the offsetting of negative margins. Moreover, the EC had not provided any evidence on the mandatory character of the Manual. Secondly, the US argued that the AD Margin Program is not a challengeable measure because it is a software program implementing the rules stemming from other instruments and is not an instrument enshrining rules itself. Assuming that it were a measure, it would not prevent the USDOC from ignoring the negative dumping margins because the Assistant Secretary of Import Administration was not obliged to follow either the Standard AD Margin Program or the Antidumping Manual. Lastly, the US considered that classification of the lines of a computer code (i.e., the standard zeroing line) as administrative procedures within the meaning of Article 18.4 AD Agreement is contrary to general rules of treaty interpretation. The US claimed that ‘practice’ is not a ‘measure’, and even if it were, it would not be mandatory within the meaning of the mandatory/discretionary test, and there is no evidence that the US treats zeroing as binding or mandatory. Therefore, on these grounds it contended that the EC's claims regarding standard zeroing procedures should be rejected.
4.1.2 US response to the EC's claims with respect to zeroing in the context of administrative reviews
4.1.2.1 US response to the EC's ‘as applied’ claims regarding the inconsistency of simple zeroing with Articles 2.4.2 and 2.4 AD Agreement
The US argued that the obligations under Article 2.4.2 AD Agreement apply only to the investigation phase of an antidumping proceeding and not to the assessment proceedings, which are a distinct phase of the antidumping proceeding and have a different purpose. It contested the application of Article 2.4.2 AD Agreement to assessment proceedings under Article 9 AD Agreement on the basis of the text of the AD Agreement and the Appellate Body and Panel reports drawing a distinction between original investigations and reviews. It argued that Article 9 AD Agreement does not contain the requirements of Article 2.4.2, and the phrase ‘the existence of margins of dumping during the investigation phase’ only applies as regards Article 5 investigations because it is the only investigation phase in the AD Agreement requiring the determination of the existence of a dumping margin. The US considered its assertion consistent with the fact that the AD Agreement provides for the use of different duty-assessment systems and the application of Article 2.4.2 AD Agreement to assessment proceedings would create the divergences between the two duty-assessment systems. Moreover, it asserted that Article 9.4 (ii) AD Agreement permits the calculation of antidumping duties by a comparison of the prospective normal value and individual export prices, and there is nothing in the AD Agreement requiring members using the prospective system to provide credit for nondumped entries while assessing duties for future entries that are dumped. It challenged the EC's claim that a limited application of Article 2.4.2 AD Agreement to the investigation phase would be disadvantageous for the retrospective-assessment-system users. Furthermore, it considered that the EC erred in arguing that the Article 5 investigation-phase margins are the basis of the duty collection in all prospective systems because in the prospective systems also different margins can be calculated in the reviews under Article 11.2 AD Agreement in order to bring the duties to the actual margin of dumping of the exporter. Article 2.4.2 is not applicable to such reviews either.
Lastly, the US argued that while the investigations determine the application of the antidumping measures, the collection and assessment of antidumping duties occur only after such measures are imposed. Moreover, since it is the importers' liability to pay the duties, a determination on importer-specific and transaction-specific basis is apt because the EC's exporter-oriented assessment process would, under any of the dumping-margin methodologies, ‘even without zeroing require some assessment of anti-dumping duties on the non-dumped imports of such an importer’.Footnote 21
As regards the Article 2.4 AD Agreement claims of the EC, the US argued that the symmetrical-comparison requirement for the export price and normal value is only dealt with in Article 2.4.2 AD Agreement, and this would be redundant if a fair-comparison obligation was already included in Article 2.4 AD Agreement. Moreover, it denied that higher duties result from the use of the average-to-transaction comparison method in comparison to the symmetrical method. The US argued that the EC's assumption that the AD Agreement requires symmetrical-comparison methods in assessment proceedings contradicts the fact that the said agreement stipulates several types of assessment systems, and, furthermore, Article 9.4(ii) AD Agreement expressly provides for the average-to-transaction comparison method for assessment purposes. Lastly, the US asserted that if the nondumped transactions are allowed to offset dumped transactions, identical results will be obtained by the application of the average-to-transaction method as by the average-to-average comparison method and the targeted-dumping provision will be rendered redundant. Moreover, the EC's argument allowing zeroing in average-to-transaction comparisons is contradictory to its challenge that zeroing is an impermissible allowance, because Article 2.4.2 AD Agreement does not suggest that a targeted-dumping provision is an exception to the fair-comparison requirement.
4.2 US response to Japan's claims in US–Zeroing (Japan)
4.2.1 US response to Japan's claims that zeroing procedures are measures subject to dispute settlement under the DSU and the AD Agreement
The US denied the existence of a standard computer program (i.e., the AD Margin Program) which could be termed as a measure within the meaning of Article 6.2 DSU and asserted that the USDOC tailors its computer programs to every antidumping investigation.Footnote 22 It argued that the standard zeroing line is just a calculation tool rather than an instrument embodying general rules having prospective application, and, even assuming that the computer programs are measures, they cannot be considered WTO-inconsistent because they do not require a mandatory offsetting of the negative margins. Furthermore, the US argued that Japan failed to identify any actual measures that correspond to the zeroing procedures, to prove how the zeroing procedures and the standard zeroing line constitute measures, and to prove their mandatory character. The US considered that the evidence of the consistent use of zeroing at best demonstrates its scope but not the existence of a measure. Lastly, the US argued that it does not admit the use of zeroing procedures or the standard zeroing line simply on the grounds of the statement it made indicating that the USDOC has never granted an offset for negative dumping. Continuing further, it stated that there are no US laws/regulations governing the calculation of dumping margins that address the issue of offsetting nondumped transactions, which is at the discretion of the USDOC Assistant Secretary.
4.2.2 US response to Japan's claims with respect to zeroing procedures in original investigations
4.2.2.1 US response to Japan's ‘as such’ claims with regard to zeroing in original investigations
4.2.2.1.1 US response to Japan's claim with respect to the inconsistency of model and simple zeroing with Articles 2.1, 2.4.2 AD Agreement and Articles VI:1, VI:2 GATT
The US argued that the AD Agreement and particularly Article 2.4.2 AD Agreement do not prescribe an obligation to calculate the dumping margin for the product as a whole, and the latter provision does not address the issue of the aggregation of the multiple comparisons for the calculation of an overall margin. It asserted that the phrase ‘product as a whole’ is not used in Article VI GATT or in Articles 2.1, 2.4, or 2.4.2 AD Agreement. Moreover, a dumping margin in terms of price difference and in tune with the usage of the term in Ad Article VI:1 GATT provides that dumping can be found in individual transactions where the export prices are less than the normal value. Hence, in the transaction-to-transaction and average-to-transaction comparison methods, the plural ‘margins of dumping’ refer to the results of multiple transaction-specific comparisons in consonance with Article VI:2 GATT. Lastly, the US alleged that Japan's argument that the second sentence of Article 2.4.2 AD Agreement is based on the assumption that the average normal value under the average-to-transaction method is established on a different basis from the average normal value in the average-to-average method lacks textual support.
4.2.2.1.2 US response to Japan's claim with respect to the inconsistency of model and simple zeroing with Article 2.4 AD Agreement
The US responded that Article 2.4 AD Agreement does not embody any obligation with respect to zeroing because the ‘fair comparison’ requirement ‘refers to adjustments necessary to account for differences between export price and normal value that affect price comparability’,Footnote 23 and this applies to the required price adjustments and not to the treatment of the results of the comparisons. Additionally, it asserted that Japan's interpretation that Article 2.4 AD Agreement contains a general obligation to offset negative dumping margins erodes the difference between the average-to-average and average-to-transaction comparison and this view is not reconcilable with the text of the AD Agreement. The US stated that Japan incorrectly argued that the USDOC Regulations require the application of the third comparison methodology to a subset of export transactions. Finally, it asserted that the Appellate Body Reports in EC–Bed Linen, US–Corrosion-Resistant Steel Sunset Review, and US–Softwood Lumber V do not provide a basis to conclude that the ‘fair comparison’ requirement in Article 2.4 AD Agreement enshrines an independent obligation to provide offsets for export transactions that exceed the normal value. In this respect, the US concluded that the simple fact that the nonuse of offsets leads to a higher dumping margin is not sufficient to conclude that the comparison is unfair.
4.2.2.1.3 US response to Japan's claim with respect to the inconsistency of model and simple zeroing with Articles 3.1 to 3.5 and 5.8 AD Agreement
The US responded that Japan's ‘as such’ claims under these provisions were unfounded because Japan failed to explain ‘how USDOC's approach necessarily results in a lack of positive evidence in any, let alone every injury determination’.Footnote 24 Arguing further, it asserted that Japan's claims as regards Article 5.8 AD Agreement are dependent upon a violation of Articles 2.1, 2.4, or 2.4.2 AD Agreement, and ‘it did not establish that, were it to prevail with respect to its claims that the United States acted in breach of Articles 2.1, 2.4 or 2.4.2, the only margins that could be determined in a WTO-consistent manner must be less than de minimis’.Footnote 25
4.2.2.2 US response to Japan's ‘as applied’ claims with respect to the use of model zeroing in the antidumping proceedings concerning cut-to-length carbon-quality steel products
The US did not provide specific arguments with regard to the use of zeroing and its consistency with Articles 2.1 and 2.4.2 AD Agreement in this antidumping proceeding.
4.2.3 US response to Japan's claims with regard to zeroing procedures in administrative reviews and new-shipper reviews
4.2.3.1 US response to Japan's ‘as such’ claims with respect to zeroing procedures in the context of administrative reviews and new-shipper reviews
4.2.3.1.1 US response to Japan's claims regarding the inconsistency of simple zeroing with Articles 2.1 and 2.4.2 AD Agreement and Articles VI:1 and VI:2 GATT
The US alleged that the AD Agreement does not require the establishment of one dumping margin for the product as a whole. It contested the Appellate Body's reasoning in US–Zeroing (EC) in this regard and asserted that Article 2.4.2 AD Agreement is only applicable to the investigation phase within the meaning of Article 5 AD Agreement. The US considered the latter assertion to be consistent with the different functions of investigations and other proceedings under the AD Agreement and with the various types of duty-assessment systems provided for. Thus, it argued that Article 9 AD Agreement contains a general reference to Article 2 including any limitation in the text of Article 2.4.2 AD Agreement, and, since the latter is limited to the investigation phase only, Article 9.3 AD Agreement does not incorporate its requirements.
4.2.3.1.2 US response to Japan's claims regarding the inconsistency of simple zeroing with Article 2.4 AD Agreement
As regards this claim, the US raised the same arguments as it did to refute Japan's arguments concerning the inconsistency of zeroing with Article 2.4 AD Agreement in the context of original investigations.
4.2.4 US response to Japan's claims with regard to zeroing procedures in the context of changed-circumstance and sunset reviews
4.2.4.1 US response to Japan's claims regarding the ‘as such’ inconsistency of zeroing with Articles 2 and 11 AD Agreement in changed-circumstances and sunset reviews
The US argued that Japan's claims against the inconsistency of zeroing ‘as such’ with respect to these provisions are untenable because the fair-comparison obligation of Article 2.4 AD Agreement cannot be understood to require offsets in all proceedings. Moreover, the Appellate Body in US–Softwood Lumber V addressed this issue only in the context of the average-to-average methodology under Article 2.4.2 AD Agreement, which by itself applies only to original investigations.Footnote 26
4.2.4.2 US response to Japan's ‘as applied’ claims in two sunset reviews
The US invoked two arguments against Japan's claims as regards the two sunset reviews. Firstly, the US denied that in the two sunset reviews at issue the USITC relied on dumping margins reported by the USDOC. Secondly, it held that Japan's claims are speculative and unfounded because it does not necessarily follow from the assumption that the dumping margins reported to the USITC are inconsistent with the AD Agreement and that the USDOC would have reported different margins had it used a different methodology.
The US argued that just as the Appellate Body has recognized, as regards Article 11.3 AD Agreement, that there is no obligation for authorities to rely on the dumping margins in making the likelihood of continuation or recurrence of dumping determinations, the same reasoning extends to the likelihood of continuation or recurrence of injury determinations as well.
5. The Reports–Panel and Appellate Body
5.1 US–Zeroing (EC)
5.1.1 Findings regarding the consistency/inconsistency of zeroing under the WTO rules
5.1.1.1 Findings with respect to model zeroing in 15 ‘as applied’ original investigations
The Panel firstly referred to the findings in EC–Bed Linen and US–Softwood Lumber V that in the calculation of the overall dumping margin for a product as a whole, using the average-to-average comparison method, the exclusion from the numerator of the weighted-average dumping margin of the results of comparisons where the average prices of all comparable export transactions exceed the average normal value violates Article 2.4.2 AD Agreement. While noting the US's arguments referring to the historical background of Article 2.4.2 AD Agreement, which were also addressed in US–Softwood Lumber V, the Panel decided not to depart from the Appellate Body's findings in this case for the reasons of ensuring the security and predictability of the multilateral trading system (as envisaged under Article 3.2 DSU).Footnote 27
Therefore, the Panel held that the US had acted inconsistently with Article 2.4.2 AD Agreement as regards the 15 original investigations at issue for the following reason:
USDOC did not include in the numerator used to calculate weighted-average dumping margins any amounts by which average export prices in individual averaging groups exceeded the average normal value for such groups.Footnote 28
5.1.1.2 Findings with respect to zeroing ‘as such’ in original investigations
5.1.1.2.1 Findings with respect to the claims concerning the provisions of the Tariff Act
The Panel clarified that legislation can be a challengeable measure ‘as such’ under the DSU independent of its application in specific instances. Pursuant to an analysis of the provisions of the Tariff Act considered WTO-inconsistent by the EC, the Panel found that those provisions do not address the methodology of calculating a dumping margin and, more specifically, they do not refer to the issue of zeroing. Therefore, it ruled that those provisions cannot be held as WTO-inconsistent for an issue that they do not address. Furthermore, the Panel reiterated the judgment of the US Court of Appeals for the Federal Circuit, which it considered had the final say regarding the meaning of the US antidumping statute. Finally, the Panel affirmed that the Tariff Act neither requires nor prohibits the USDOC from applying zeroing. Therefore, it held that since the challenged provisions cannot be considered mandatory within the meaning of the mandatory/discretionary test, hence, they cannot be considered as WTO-inconsistent ‘as such’. The Panel concluded that:
Sections 771(35)(A) and (B), 731 and 777A(d) of the Tariff Act are not as such inconsistent with Articles 2.4, 2.4.2, 5.8, 9.3, 1 and 18.4 of the AD Agreement, Articles VI:1 and VI:2 of the GATT 1994 and Article XVI:4 of the WTO Agreement with respect to the use of a zeroing methodology in the calculation of margins of dumping in original investigations.Footnote 29
5.1.1.2.2 Findings with respect to the claims concerning standard zeroing procedures
As regards the EC's claims concerning the practice of zeroing and the standard zeroing procedures, the Panel while considering the Appellate Body findings in US–Corrosion-Resistant Steel Sunset Review and US–Oil Country Tabular Goods Sunset Reviews Footnote 30 held that it is possible to challenge a measure as an act setting rules for general/prospective application even if it is not a legal instrument and does not have a binding character. In this context, the Panel observed that the standard zeroing procedures are used in a specific antidumping proceeding by virtue of their incorporation in the particular computer program used by the USDOC in that proceeding. Therefore, they cannot be considered as acts/instruments intended to have general/prospective application on their own, and accordingly, they cannot be considered administrative procedures within the meaning of Article 18.4 AD Agreement. The Panel also evaluated the EC's contention regarding the existence of the practice of zeroing ‘as such’ and its WTO-consistency, because it held that it is possible to challenge a norm as a measure even if such a norm does not exist in an officially written form but its existence is manifest on the basis of other evidence. Therefore, on the basis of the evidence before it regarding the constant inclusion of the standard zeroing procedures in the computer programs for an extended period of time by the USDOC and the absence of any instances of giving credit for nondumped sales, the Panel held that the zeroing is a well-defined norm of the USDOC. Moreover, it held that since model zeroing in original investigations is inconsistent with Article 2.4.2 AD Agreement, the USDOC's practice of zeroing ‘as such’ will lead to WTO-inconsistent actions. Thus, the Panel concluded that:
the United States' zeroing methodology, as it relates to original investigations, is a norm which, as such, is inconsistent with Article 2.4.2 of the AD Agreement.Footnote 31
In view of this finding, the Panel considered it irrelevant to consider EC's claims under other provisions of the AD Agreement, GATT, and WTO Agreement.
The US appealed that the Panel had erred in ruling that the zeroing methodology is a measure challengeable ‘as such’. The Appellate Body rejected the US's claim and upheld the Panel findings that the use of the zeroing while calculating the overall dumping margin using the average-to-average comparison method in original proceedings can be challenged ‘as such’ in a WTO proceeding.
5.1.1.3 Findings with respect to simple zeroing ‘as applied’ in 16 administrative reviews
5.1.1.3.1 Findings with respect to the claims under Article 2.4.2 and 2.4 AD Agreement
Deliberating on all the assertions of the EC in the context of its claim that Article 2.4.2 AD Agreement applies to any proceeding under the AD Agreement, the Panel firstly held that the phrase ‘existence of margins of dumping during the investigation phase’ in Article 2.4.2 applies only to original investigations within the meaning of Article 5 AD Agreement. Of the several reasons for its finding, the Panel considered that the terms ‘investigation’ and ‘investigations’ mentioned in the different provisions of the AD Agreement refer particularly to the investigation proceedings as envisaged under Article 5.1 and are not used in relation to proceedings that take place once an antidumping measure has been adopted. Moreover, it considered that there is nothing in Articles 9 and 11 AD Agreement that indicates that proceedings under these Articles are ‘investigations’. While noting the distinction made between investigations and subsequent proceedings in various Appellate Body decisions, the Panel also underlined the difference drawn by Article 18.3 AD Agreement between ‘investigations’ and ‘reviews of existing measures’.
Secondly, the Panel also rejected the EC's claim that limiting the application of Article 2.4.2 AD Agreement to original investigations would be contrary to Article 9.3 AD Agreement and would lead to unequal treatment of prospective and retrospective duty-assessment systems. To begin with, it considered that rules regarding the imposition and collection of the antidumping duties under Article 9 AD Agreement are distinct from the rules on the determination of the dumping margin. It held that the former focuses on the overall behavior of the importer and the latter translates into an importer-specific, transaction-specific duty liability, and Article 9.3 AD Agreement does not require an exporter-oriented assessment as argued by the EC. It continued to observe that there is an inherent difference between the prospective and retrospective duty systems regardless of the application of Article 2.4.2 AD Agreement, and the AD Agreement, as such, cannot be interpreted to mean that both the duty-assessment systems would lead to the same level of protection against dumped imports.
Moreover, rejecting the EC's arguments regarding the subsequent practice of WTO members and supplementary means of treaty interpretation, the Panel held that Article 2.4.2 AD Agreement applies only to original investigations within the meaning of Article 5 AD Agreement and not to the duty-assessment proceedings under Article 9.3 AD Agreement.
With respect to the EC's claims under Article 2.4 AD Agreement, the Panel held that though there is a close connection between the first sentence and the remainder of the paragraph of Article 2.4 AD Agreement, ‘it does not suggest that the fair comparison requirement is defined exhaustively by the specific requirements set out in the remainder of the paragraph regarding steps to be taken to ensure price comparability’.Footnote 32 The Panel considered that the fair-comparison obligation is not limited to ensuring price comparability by selecting comparable transactions or through appropriate adjustments but also applies to the calculation of dumping margins under Article 2.4.2 AD Agreement. It opined that for determining what is ‘fair’ under the AD Agreement as regards dumping-margin calculations, Articles 2.4.2 and 9 AD Agreement have to be considered because the former Article is the only provision addressing the methods of calculating dumping margins. On an analysis of these provisions, the Panel held that interpreting Article 2.4 AD Agreement as prohibiting zeroing and asymmetrical comparisons in original investigations and in duty-assessment proceedings would render Article 2.4.2 AD Agreement ineffective. It noted that there is no Appellate Body decision that considers zeroing inconsistent with Article 2.4 AD Agreement.
Furthermore, the Panel also rejected the EC's assertion that zeroing is an impermissible allowance for a difference not affecting price comparability and is inconsistent with the third to fifth sentences of Article 2.4 AD Agreement. It held that this argument cannot be reconciled with the fact that the second sentence of Article 2.4.2 AD Agreement addresses zeroing only as regards original investigations and the asymmetrical comparisons permitted by this provision would be rendered nugatory if zeroing is prohibited. On this basis, the Panel ruled that Article 2.4 AD Agreement does not proscribe the use of simple zeroing in the average-to-transaction method in assessment proceedings under Article 9.3 AD Agreement.
Therefore, the Panel held that the US did not violate Articles 2.4.2 and 2.4 AD Agreement in the 16 administrative reviews at issue by using the asymmetrical average-to-transaction comparison method to compare the normal value and export price and by excluding any amounts by which the export prices exceeded the normal values. Moreover, since the Panel rejected the EC's claims under Articles 2.4.2 and 2.4 AD Agreement, it also rejected the dependent claims under Articles 1, 9.3, 11.1, 11.2, 18.4 AD Agreement; Articles VI:1, VI:2 GATT; and Article XVI:4 WTO Agreement.
The EC appealed the Panel's findings and argued that simple zeroing in administrative reviews is WTO-inconsistent. In this regard, the Appellate Body analyzed the term ‘margins of dumping’ used in Article 9.3 AD Agreement and Article VI:2 GATT and noted that the former provision refers to the dumping margins established under Article 2 AD Agreement. It reiterated the findings made in EC–Bed Linen and US–Softwood Lumber V and particularly referred to the latter dispute where it was ruled that dumping is defined in relation to the product as a whole and though multiple comparisons at an intermediate stage are permissible, the dumping margin is to be established for the product as a whole by aggregating all the intermediate values. Therefore, the inclusion of some comparisons and the exclusion of others while calculating the dumping margin using the average-to-average method is not justified. Accordingly, the Appellate Body applied the above-mentioned reasoning with respect to Article 9.3 AD Agreement and ruled that the dumping margin for an exporter limits the maximum antidumping duty that can be levied on the entries of the subject product from that exporter. It explained that this means that a comparison has to be made between the antidumping duties collected on all the entries of the subject product from an exporter with that exporter's dumping margin calculated for the product as a whole. The Appellate Body evaluated the US's assessment method whereby the individual transactions where the export prices exceeded the average normal values were disregarded leading to the assessment of duty amounts in excess of the exporter's dumping margin (with which the duties had actually to be compared as per Article 9.3 AD Agreement and Article VI:2 GATT). Thus, the Appellate Body partly reversed the Panel findings and held that by using simple zeroing in the administrative reviews, the USDOC had violated Article 9.3 AD Agreement and Article VI:2 GATT. Furthermore, having reversed the Panel findings as regards Article 9.3 AD Agreement, the Appellate Body declared the Panel ruling that the US had not infringed the fair-comparison obligation under the first sentence of Article 2.4 AD Agreement by using simple zeroing in the administrative reviews as moot.
The EC had also appealed that the Panel had wrongly rejected its argument that by applying simple zeroing the US makes an allowance for a difference not affecting price comparability and thus acts inconsistently with the third to fifth sentences of Article 2.4 AD Agreement. The Appellate Body analyzed that if allowances which do not affect price comparability are made, the intent of the requirement under the third sentence of Article 2.4 AD Agreement would be eclipsed. Applying this interpretation a contrario, it considered that allowances for differences that do not affect price comparability should not be made. Therefore, it held that an allowance like zeroing made in relation to price differences between the export and domestic transactions is not impermissible under the third to fifth sentences of Article 2.4 AD Agreement and accordingly upheld the Panel findings in this regard.Footnote 33
Lastly, the Appellate Body upheld the Panel's rejection of the EC's dependent claims under Articles 11.1 and 11.2 AD Agreement.
5.1.1.4 Findings with respect to the ‘as such’ claims concerning zeroing in administrative reviews
The Panel rejected the EC's dependent claims regarding the maintenance of standard zeroing procedures; the practice of zeroing; and sections of the Tariff Act and USDOC Regulations in the context of administrative reviews on account of their alleged inconsistency with provisions of the AD Agreement, GATT, and WTO Agreement because it had rejected the independent claims under Articles 2.4 and 2.4.2 AD Agreement.
On appeal by the EC, the Appellate Body held that since these claims were dependent on the Panel findings with regard to Articles 2.4 and/or 2.4.2 AD Agreement that it had partly reversed (i.e., in context of Article 9.3 AD Agreement and Article VI:2 GATT) and partly declared moot (i.e., as regards the first sentence of Article 2.4 AD Agreement), it declared the Panel findings as regards the consistency of zeroing with the alleged provisions of the AD agreement, GATT, and WTO Agreement also moot. Additionally, the Appellate Body also declared the Panel findings concerning section 351.414(c)(2) USDOC Regulations and the zeroing methodology that were not found to be inconsistent ‘as such’ also moot and declined to complete the analysis on both the issues.
5.1.1.5 Findings with respect to the ‘as such’ claims concerning zeroing in new-shipper reviews, changed-circumstances reviews, and sunset reviews
The Panel held that the standard zeroing procedures; sections 771(35)(A) and (B), 731, 777A(d), and 715(a)(2)(i) and (ii) of the Tariff Act; and section 351.414(c)(2) USDOC Regulations in the context of new-shipper reviews, changed-circumstances reviews, and sunset reviews do not ‘as such’ infringe Articles 2.4, 2.4.2, 9.3, 9.5, 11.1, 11.2, 11.3, 1, 18.4 AD Agreement, Articles VI:1, VI:2 GATT, and Article XVI:4 WTO Agreement on the grounds that these claims were dependent on a violation of Articles 2.4 and 2.4.2 AD Agreement that it had rejected.
5.2 US–Zeroing (Japan)
5.2.1 Findings with regard to the consistency/inconsistency of zeroing under the WTO rules
5.2.1.1 Findings on whether zeroing procedures constitute measures ‘as such’ subject to the dispute settlement under the DSU and the AD Agreement
With respect to Japan's claims that the zeroing procedures and the standard zeroing line constitute measures that are challengeable ‘as such’ under the WTO dispute settlement, the Panel reviewed the Appellate Body findings in the recent casesFootnote 34 and evaluated whether the zeroing procedures and the standard zeroing line have a precise identifiable content; are attributable to the US; and concern rules or norms intended to have general and prospective application.
Accordingly, the Panel adjudicated that the standard zeroing line is not a measure challengeable ‘as such’ because it is merely an instruction in a computer program relating to a particular aspect of a dumping-margin calculation and is not a norm/standard of general application on its own because it has to be specifically included in each computer program used in a particular investigation or review. As regards the standard zeroing procedures, the Panel considered it to imply the zeroing methodology per se and held it to be a measure challengeable ‘as such’. It based this conclusion firstly on the evidence which proved that zeroing has been consistently used by the USDOC for some time, and the standard zeroing line has been incorporated by the US in a majority of the computer programs, and, when it is not incorporated, other methods are used for zeroing nondumped transactions. Secondly, it relied on the repeated statements of the USDOC and other US agencies confirming the deliberate and consistent application of zeroing as a rule of general and prospective application regardless of the comparison method used. The Panel also clarified that model and simple zeroing were just two different expressions of a single rule.Footnote 35
On appeal, the Appellate Body upheld the Panel findings in this respect and rejected the US's claim that the Panel had infringed its objective-assessment obligation under Article 11 DSU. It affirmed the Panel findings that zeroing constitutes a challengeable measure ‘as such’ to the extent that it relates to the calculation of dumping margins on the basis of transaction-to-transaction and average-to-transaction comparisons in original investigations.
5.2.1.2 Findings with respect to the claims regarding model and simple zeroing in original investigations
5.2.1.2.1 Findings with respect to the ‘as such’ and ‘as applied’ claims concerning model zeroing with regard to Articles 2.1, 2.4.2 AD Agreement, and Articles VI:1, VI:2 GATT
The Panel held that the use of model zeroing by the USDOC in the context of original investigations is ‘as such’ inconsistent with Article 2.4.2 AD Agreement because the dumping margin so calculated does not take into account all comparisons between the normal value and the export price. It reasoned further that the first sentence of Article 2.4.2 AD Agreement proscribes model zeroing because it requires the comparison between the weighted-average normal value and weighted-average export price reflecting the prices of all comparable export transactions, and text of the said Article does not indicate that dumping margins can be determined for individual models.
Exercising judicial economy with respect to Japan's request to make additional findings under Articles 2.1 and 2.4 AD Agreement and Articles VI:1 and VI:2 GATT, the Panel did not consider it necessary to rule further because it had declared model zeroing inconsistent with Article 2.4.2 AD Agreement. The Panel also held that by applying model zeroing in the antidumping investigation of imports of certain cut-to-length carbon-quality steel products from Japan, the US had infringed Article 2.4.2 AD Agreement.
5.2.1.2.2 Findings with respect to the ‘as such’ claims concerning simple zeroing with regard to Articles 2.1, 2.4.2 AD Agreement, and Articles VI:1, VI:2 GATT
The Panel firstly clarified that Japan's extension of the Appellate Body's findings in US–Softwood Lumber V to the context of simple zeroing is not acceptable because the said dispute dealt with multiple averaging and the Appellate Body's reasoning cannot be extended beyond that context. Furthermore, on the basis of a textual analysis of the words ‘product’ and ‘products’ in Articles 2.1, 2.4.2 AD Agreement, and Article VI:1, VI:2 GATT, the Panel considered that the said provisions do not imply a general requirement to determine ‘dumping’ and dumping margins for the product as a whole, and none of these provisions contain the phrase ‘product as a whole’. Moreover, concurring with the Panel in US–Softwood Lumber V (Article 21.5),Footnote 36 it held that the use of the word ‘product’ in these provisions does not preclude the possibility of establishing a dumping margin on a transaction-specific basis. Accordingly, the Panel concluded as follows:
The fact that the terms ‘dumping’ and ‘margin of dumping’ in Article 2.1 of the AD Agreement and Articles VI:1 and VI:2 of the GATT are defined in relation to ‘product’ and ‘products’ does not warrant the conclusion that these terms, by definition, cannot apply to individual transactions and inherently require an examination of export transactions at an aggregate level in which the same weight is accorded to export prices that are above normal value as to export prices that are below normal value.Footnote 37
Furthermore, analyzing the consistency of simple zeroing with Article 2.4.2 AD Agreement, the Panel considered that this provision on the one hand does not generally prohibit zeroing and on the other hand permits transaction-to-transaction comparisons. The Panel considered this to imply that the positively dumped transactions may be considered more relevant by a member than the negatively dumped transactions. It extended the same reasoning with respect to the average-to-transaction method. The Panel also considered it impossible to reconcile the general prohibition of zeroing with the second sentence of Article 2.4.2 AD Agreement because, in this situation, the results obtained from application of the average-to-average method would be identical to those obtained under the average-to-transaction method thus rendering the second sentence of the said provision inutile. The Panel considered it untenable that Article 2.4.2 AD Agreement permits zeroing in the average-to-transaction method while prohibiting it under the other two methods. Therefore, it concluded that simple zeroing applied by the USDOC in original investigations is permissible and does not ‘as such’ infringe Articles 2.1 and 2.4.2 AD Agreement and Articles VI:1, VI:2 GATT.
On appeal, the Appellate Body followed the US–Softwood Lumber V (Article 21.5) Footnote 38 ruling that zeroing while using the transaction-to-transaction comparison method in original investigations is inconsistent with Article 2.4.2 AD Agreement. It considered that the absence of the words ‘all comparable transactions’ in the context of the transaction-to-transaction method in Article 2.4.2 AD Agreement does not mean that zeroing is permitted. The Appellate Body considered further that since both the average-to-average method and the transaction-to-transaction method have the same purpose (i.e., calculation of dumping margins), it would be illogical to interpret the latter in a manner that leads to results different from the former. It disagreed with the Panel that ‘dumping’ can be determined for individual transactions and multiple-comparison results are dumping margins in themselves; and that as per the first sentence of Article 2.4.2 AD Agreement, dumping margin means the total amount by which the transactions-specific export prices are less than transactions-specific normal values. Furthermore, the Appellate Body held that the second sentence of Article 2.4.2 AD Agreement addressing targeted dumping is an exception to the normal methods. It rejected the Panel's assumption that a general prohibition of zeroing would lead to mathematically equivalent results under the first and second sentences of Article 2.4.2 AD Agreement because an investigating authority is obliged to aggregate the results of all transaction-specific comparisons and the export transactions would be more limited in the second sentence (as they would be limited to the ones falling within the particular pricing pattern). The Appellate Body concluded as follows:
In the light of our analysis of Article 2.4.2 of the Anti-Dumping Agreement, we conclude that, in establishing ‘margins of dumping’ under the T-T comparison methodology, an investigating authority must aggregate the results of all the transaction-specific comparisons and cannot disregard the results of comparisons in which export prices are above normal value.Footnote 39
Therefore, it reversed the Panel findings and held that the USDOC violated Article 2.4.2 AD Agreement by using zeroing in the transaction-to-transaction comparison method in original investigations.
The Appellate Body also reversed the Panel's findings in relation to Article 2.1 AD Agreement and Articles VI:1 and VI:2 GATT since they were based on its findings concerning Article 2.4.2 AD Agreement.
5.2.1.2.3 Findings with respect to the ‘as such’ claims regarding simple zeroing in the context of Article 2.4 AD Agreement
The Panel firstly held that the fair-comparison obligation in the first sentence of Article 2.4 AD Agreement is an independent obligation and is not limited to the issue of allowances to ensure price comparability because it is not defined by the remaining part of Article 2.4 AD Agreement. Secondly, it noted that there is no Appellate Body ruling rendering zeroing inconsistent with Article 2.4 AD Agreement on its own. Furthermore, concurring with the Panel in US–Zeroing (EC), it held that the fair-comparison requirement cannot be interpreted to generally prohibit zeroing. Therefore, the Panel concluded that the USDOC does not act inconsistently with Article 2.4 AD Agreement by maintaining simple zeroing in original investigations.
On appeal, the Appellate Body disagreed with the implication of the Panel's reasoning that the fair-comparison requirement was dependent on Article 2.4.2 AD Agreement because the latter provision constituted a lex specialis vis-à-vis Article 2.4 AD Agreement. Moreover, the Appellate Body reiterated the findings made in US–Softwood Lumber V that zeroing in the transaction-to-transaction comparison method in original investigations inflates the magnitude of dumping by artificially reducing the prices of certain export transactions and makes positive dumping determinations more likely. Thus, it held that this method of dumping-margin calculation is not unbiased or evenhanded and, accordingly, zeroing in transaction-to-transaction comparisons violates the fair-comparison requirement. Consequentially, the Appellate Body reversed the Panel's decision in this regard and held that the US infringed Article 2.4 AD Agreement by maintaining simple zeroing in original investigations.
5.2.1.2.4 Findings concerning the ‘as such’ claims regarding zeroing in the context of Articles 3.1 to 3.5, 5.8, 1, 18.4 AD Agreement and Article XVI:4 WTO Agreement
The Panel held that since simple zeroing is not inconsistent with Articles 2.1, 2.4.2 AD Agreement and Articles VI:1, VI:2 GATT, the USDOC does not violate the above-captioned provisions by maintaining simple zeroing in original investigations.
5.2.1.3 Findings with respect to the ‘as such’ claims regarding model and simple zeroing in the context of administrative reviews and new-shipper reviews
The Panel applied the same line of reasoning to reject Japan's ‘as such’ claims with respect to the inconsistency of zeroing in the context of administrative reviews and new-shipper reviews with Articles 2.1, 2.4.2, 2.4, 9.1–9.3, and 9.5 AD Agreement and Articles VI:1 and VI:2 GATT, as it did in the context of original investigations. In the light of the express reference to a prospective normal value system in Article 9.4(ii) AD Agreement, the Panel concluded that the AD Agreement and the GATT provisions neither require dumping and dumping-margin determinations to be established for the product as a whole nor do they prohibit zeroing. It considered that these two Agreements do not warrant an aggregate examination of nondumped and dumped export transactions equally, for determining the existence of dumping. The Panel also considered that Article 9.3 AD Agreement requires the limiting of the antidumping-duty amount to the dumping margin established under Article 2 AD Agreement, and the obligation to pay an antidumping duty is on the importer on an import-specific basis. Therefore, the latter obligation would be violated if the dumping margin under Article 9.3 AD Agreement is calculated by an aggregation of export prices during a review period in which export prices above the normal value are treated equally as the export prices below the normal value. The implication of this was noted by the Panel that in a retrospective duty-assessment system, a member may be precluded from collecting antidumping duties in respect of the lower-than-normal-value export transactions of a particular importer at a particular point of time because the prices of export transactions to other importers exceed normal value at a different point in time. As regards the Article 2.4.2 AD Agreement arguments invoked by Japan, the Panel emphasized that this provision does not support the fact that zeroing is prohibited under the AD Agreement or GATT in the context of average-to-average comparisons because a contrary interpretation would make the average-to-transaction method redundant. It held that Article 2.4.2 AD Agreement is applicable only to original investigations under Article 5 AD Agreement and concurred with the US–Zeroing (EC) Panel in this regard.
Secondly, the Panel determined that simple zeroing in administrative and new-shipper reviews does not violate Article 2.4 AD Agreement because interpreting the fair-comparison obligation of this provision as meaning a general prohibition of zeroing would render other AD Agreement provisions ineffective. Moreover, the interpretation that zeroing is prohibited under any comparison method and any proceeding is not supported by Article 9 AD Agreement. Therefore, the Panel concluded that by maintaining simple zeroing in administrative and new-shipper reviews, the US does not violate Articles 2.1, 2.4.2, 2.4 AD Agreement and Articles VI:1 and VI:2 GATT.
In the light of the above findings, the Panel rejected Japan's consequential claims under Articles 1, 9.1–9.3, 9.5, and 18.4 AD Agreement and Article XVI:4 GATT and also concluded that the USDOC did not act inconsistently with Articles 1, 2.1, 2.4.2, 2.4, and 9.1–9.3 AD Agreement and Articles VI:1 and VI:2 GATT by applying simple zeroing in the 11 administrative reviews.
On appeal, the Appellate Body reversed the Panel findings and held that the US violates Articles 9.3 and 9.5 AD Agreement and Article VI:2 GATT by maintaining simple zeroing in administrative and new-shipper reviews. The Appellate Body reasoned its decision on the grounds that dumping and dumping margins can only exist at the level of a product, and the dumping margin acts as a ceiling for the total amount of antidumping duties that can be collected in both duty-assessment systems.
Furthermore, the Appellate Body reversed the Panel findings that zeroing in administrative and new-shipper reviews is not inconsistent with Article 2.4 AD Agreement because it held that zeroing leads to the collection of antidumping duties in excess of the dumping margin, which contravenes the fair-comparison obligation enshrined in Article 2.4 AD Agreement. Consequentially, it held that by maintaining simple zeroing in administrative and new-shipper reviews, the USDOC violates Articles 2.1, 9.1, and 9.2 AD Agreement and Article VI:1 GATT and had acted in contravention of these provisions by applying simple zeroing in the 11 administrative reviews in question.
5.2.1.4 Findings with respect to the ‘as such’ claims regarding zeroing procedures in the context of changed-circumstances reviews and sunset reviews and ‘as applied’ claims concerning the two sunset reviews
The Panel first noted that Articles 11.2 and 11.3 AD Agreement are silent on the point whether the investigating authorities are required to calculate dumping margins in the changed-circumstances reviews and sunset reviews. Furthermore, reiterating the Appellate Body's ruling in US–Corrosion-Resistant Steel Sunset Reviews, the Panel held that in case the investigating authorities rely on previously calculated margins, these should be in conformity with Article 2.4 AD Agreement. Noting that Japan had not provided evidence to establish that a rule of prospective application exists, which requires the USDOC to rely on dumping margins calculated in prior proceedings to support its determinations, the Panel concluded that Japan had failed to make a prima facie case with regard to this claim.
With respect to the ‘as applied’ claims concerning the two sunset reviews, the Panel held that the USDOC and the USITC did not infringe Articles 2 and 11 AD Agreement by relying on dumping margins calculated in previous proceedings. The Panel based its ruling firstly on Japan's failure to adduce evidence in support of its assertion that in these two cases the USDOC and USITC had actually relied on the previously calculated dumping margins and secondly on the fact that the USDOC had relied on historical dumping margins but those were calculated in previous administrative reviews. The Panel held that with regard to administrative reviews it had held that the AD Agreement does not prohibit zeroing; therefore, the US did not infringe Articles 2 and 11 AD Agreement.
On appeal, the Appellate Body reversed the Panel findings. It recalled that the presence of the terms ‘review’ and ‘determine’ in Article 11.3 AD Agreement require a reasoned conclusion based on positive evidence and sufficient factual basis in sunset reviews; therefore, if the authorities relied on historical dumping margins, these margins should be in conformity with Article 2.4 AD Agreement. Hence, the Appellate Body held that since it had concluded earlier that zeroing in administrative reviews is inconsistent with Articles 2.4 and 9.3 AD Agreement, therefore by relying in the two sunset reviews on the dumping margins calculated in the administrative reviews using zeroing, the US had infringed Article 11.3 AD Agreement.
6. Evaluating the key legal and economic issues and methodologies raised by the disputes
6.1 What is zeroing?
Article 2 of the AD Agreement establishes the parameters for determining the existence and extent of dumping, which is defined in Article 2.1 as occurring when the export price of the product exported from one country to another is less than the comparable price, in the ordinary course of trade, for the like product when destined for consumption in the exporting country (the normal value). In reaching a conclusion about the existence of dumping, Article 2.4 lays down the principle that a fair comparison shall be made between the export price and the normal value. Article 2.4.2 AD Agreement provides further details on how comparisons may be made, distinguishing between three methods:
Subject to the provisions governing fair comparison in paragraph 4, the existence of margins of dumping during the investigation phase shall normally be established on the basis of a comparison of a weighted average normal value with a weighted average of prices of all comparable export transactions or by a comparison of normal value and export prices on a transaction-to-transaction basis. A normal value established on a weighted average basis may be compared to prices of individual export transactions if the authorities find a pattern of export prices which differ significantly among different purchasers, regions or time periods, and if an explanation is provided as to why such differences cannot be taken into account appropriately by the use of a weighted average-to-weighted average or transaction-to-transaction comparison.
Article 2.4.2 therefore expresses a preference for comparing normal value and export prices on a ‘weighted average-to-weighted average’ basis or on a ‘transaction-to-transaction’ basis, but allows for a third method – comparison of a weighted-average normal value to individual export transactions – if particular conditions exist.
Zeroing refers to the practice, conducted in some jurisdictions, of replacing the actual amount of dumping calculated for model or individual sales comparisons that yield negative dumping margins (i.e., models or export transactions for which the export price exceeds the calculated normal value) with a value of zero prior to the final calculation of a weighted-average margin of dumping for the product under investigation with respect to the exporters under investigation. Because the zeroing method drops transactions that have negative margins, it has the effect of increasing the overall dumping margins.
6.1.1 Zeroing under various comparison methods: simple zeroing
The examples below will clarify the various zeroing issuesFootnote 40 further. We assume that the data in Table 1 represent net prices for separate transactions on a series of dates in the month of September.Footnote 41 To keep the example as simple as possible, we will assume that each transaction is for the same volume, i.e. one unit. The USDOC computes dumping margins on a weighted-average basis, but for the purposes of our illustration, the introduction of different quantities on different dates just serves to complicate the computations – and needless complication is a primary reason why antidumping is so misunderstood.
Table 1. Sales-data example ($ per unit)
6.1.1.1 Transaction-to-transaction method
As demonstrated, our example contains nine sales in the export market and 12 sales in the home market. Under the transaction-to-transaction method, the first step of the USDOC's computations would be to match the export transactions with the comparable home-market sales made on or at about the same date. Transposing this into our example, therefore, the dumping margin would be computed on the basis of the comparison on the nine dates between the export and the home-market transactions. As is often the case in the real world, on some dates the export price is below the home-market price, on others the export price is above the home-market price, and occasionally the same price is charged in both the markets.
6.1.1.1.1 Without zeroing
Under this method, administrative authorities like the USDOC begin by calculating the difference in price on a transaction-by-transaction basis and then compute the weighted average of these price differences, i.e. the individual export transactions are compared with the individual domestic transactions made at or at about the same date as the export transactions concerned. Table 2 illustrates how the method works. For convenience, we reproduce the key sales information in columns (1)–(3). In column (4) of Table 2, we compute the difference for each comparable transaction. Accordingly, for some comparisons the difference is positive (which means dumping) and for other comparisons it is negative. When we sum the (weighted) price differences, we find that for all comparable transactions the cumulative difference is zero. Said differently, the dumping amount (35) for the two transactions with positive dumping is exactly equal to the amount (–35) for the five transactions with negative dumping. In this example, as long as the dumped and the nondumped export transactions are allowed to offset each other, the conclusion using the transaction-to-transaction method will be that there is zero dumping.
Table 2. Simple zeroing
The use of the transaction-to-transaction method is relatively rare, but authorities may decide to use it if there are few export transactions (or if they cannot employ model zeroing any more).Footnote 42
6.1.1.1.2 With zeroing
As clean and simple as the above calculations are, the USDOC has had a long practice of not computing the margins as described. Instead, in the process of the transaction-to-transaction comparisons the USDOC would employ the practice of zeroing. This practice eventually became known as simple zeroingFootnote 43 in order to distinguish it from model zeroing (a later development that we will discuss below).
In our example, and in fact in most ‘real world’ cases, the use of zeroing leads to dramatically different margins. To see this, in column (6) of Table 2 we have computed the difference for each comparable transaction using zeroing. In our example, the amount of dumping is 35, which implies a dumping margin of 3.9% (35 divided by the total export value of 900=0.039).Footnote 44
6.1.1.2 Weighted-average-to-transaction method
6.1.1.2.1 Without zeroing
The weighted-average-to-transaction method is another method of calculating margins. However, it is considered as the exceptional method under Article 2.4.2 AD Agreement, and can be used only:
if the authorities find a pattern of export prices which differ significantly among different purchasers, regions or time periods, and if an explanation is provided as to why such differences cannot be taken into account appropriately by the use of a weighted average-to-weighted average or transaction-to-transaction comparison.
Under this method, the average home-market price (which in our example is 100) is compared to the prices of individual export transactions. The comparisons are shown in column (5) of Table 2. In this case, the overall dumping margin would be zero because the positive dumping amounts calculated for the first three export sales, totaling 55, are completely offset by the negative dumping amounts in the last five export sales, totaling −55.
6.1.1.2.2 With zeroing
Prior to the entry into force of the AD Agreement in 1995, many authorities would routinely use the exceptional method to calculate dumping margins. They would then take the position that the export transactions generating negative margins are not dumped and would subsequently replace the negative dumping amounts with zeros.Footnote 45
This practice was unsuccessfully challenged in the GATT on several occasions. However, pressure from countries such as Japan, Korea, Singapore, and Hong Kong led to the new Article 2.4.2 during the Uruguay Round negotiations, which relegated the ‘asymmetrical’ method itself to an exception.
The result of zeroing with the weighted-average-to-transaction method is given in column (7) of Table 2. In the example above, the same set of objective sales data that yields a 0% margin without zeroing would produce a 6.1% dumping margin with zeroing.Footnote 46
6.1.2 Zeroing under various comparison methods: model zeroing
6.1.2.1 Weighted-average-to-weighted-average method
Perhaps the simplest way of calculating a dumping margin is via the weighted-average-to-weighted-average method. With this method, the weighted-average normal value is compared with the weighted-average export price to determine if dumping exists. Looking again to Table 2 we begin by calculating the total value of the comparable sales for the export (column 2) and the home-market (column 3) transactions ($900 for each) and divide each by the total sales (nine units in each market) to get a weighted-average price of $100 in each market. These calculations are reported at the bottom of columns (2) and (3) in Table 2. Under the weighted-average-to-weighted-average method, we conclude that there is no dumping. In the weighted-averaging process in each market, ‘high’ prices and ‘low’ prices are averaged; therefore, zeroing would normally not be an issue unless we allow for some further complication, like introducing the notion of ‘models’.
One might think that the simplicity and transparency of the weighted-average-to-weighted-average method would make it a popular method for calculating margins. However, prior to 1995 the USDOC and other users of the AD instrument generally calculated dumping margins using the weighted-average-to-transaction method. And, as the above example demonstrates, this method not only allowed the USDOC to compare a single weighted-average normal value with individual transactions, but when used along with zeroing, it allowed the USDOC to drop higher-priced export transactions.
Virtually as soon as the AD Agreement entered into force on 1 January 1995, the USDOC (and the EU Commission) adopted a variation of the simple zeroing technique that had been used in the past. As we have already discussed, Article 2.4.2 AD Agreement mandates the use of either the weighted-average-to-weighted-average method or the transaction-to-transaction method, but under specific, exceptional circumstances allows for use of the weighted-average-to-transaction method. Thus, while not specifically precluding or sanctioning the practice of zeroing, Article 2.4.2 seemingly limited the scope for resort to zeroing since the method where it seems most likely allowed – the weighted-average-to-transaction methodology – would be exceptional.
However, rather than using simple zeroing under the exceptional method, the USDOC adopted a new margin-calculation technique – the so-called model zeroing. Model zeroing refined the standard calculations by subdividing the comparable product into models and making price comparisons on a model-by-modelFootnote 47 basis in a first stage before weight averaging the results of these comparisons to produce a result for the product under investigation. Thus, after the first stage (the model-by-model comparisons) a positive or negative dumping amount will result for each model. Model zeroing is the practice of then imputing a value of zero to the model comparisons that generate negative dumping margins, preventing the results of those comparisons from offsetting the effects of models found to be positively dumped.
There is a distinct difference between the product under investigation and the concept of a ‘model’. The product concerned in most antidumping disputes is described fairly broadly – ‘Cold-rolled steel’, ‘Welded steel pipe and tubes’, ‘DRAMs’, ‘Live swine’, ‘Bed linen’, etc. Typically, each product involves multiple harmonized-system tariff-line items (referred to as HS codes). For example, cold-rolled steel might have an HS code designating ‘thickness less than 1 mm’, another designating ‘thickness between 1 mm and 3 mm’, another designating ‘thickness between 3 mm and 8 mm’, etc.
In very rare investigations, a model may correspond to a specific HS code. A second and by far the most common way for the USDOC and other administering authorities to define a model is to focus on the subcategories of an HS code. For example, if the HS code defines a range of possible thicknesses, a model might specify a particular thickness of cold-rolled sheet or perhaps a narrower range of thicknesses. In other investigations, the USDOC's definition of a model may span several HS codes. For example, a model might specify a particular finish applied to the steel – say, painted sheet – that does not correspond to a single HS code but rather overlaps multiple codes. Alternatively, authorities may establish their own parameters for defining models, sometimes with input from the complaining domestic producers.
Two important ideas emerge here. First, there is no set rule as to how the USDOC or other authorities define models. This makes it almost impossible for an exporter to know in advance how its products' weighted-average prices will be computed. Second, regardless of exactly how the model is defined, authorities perform their dumping calculations initially at the model level.Footnote 48
To understand the application of the model zeroing method, we extend our previous example and subdivide the single like product into three distinct models, A, B, and C (see column (8) in Table 3). Model A was sold on 2, 4, and 18 September; Model B was sold on 8, 10, and 16 September; and Model C was sold on 12, 20, and 24 September.
Table 3. Model zeroing
6.1.2.1.1 Without zeroing
Under this method, the first step involves the calculation of the weighted-average prices for both the export and the home-market transactions for each model. These are reported in columns (9) and (10) of Table 3. As shown in column (11) for model A (three transactions) the amount of dumping is 10, for model B (three transactions) the amount of dumping is −2, and for model C (three transactions) the amount of dumping is −8. The second step involves the aggregation of the margins across the models. Without zeroing, the USDOC would simply sum across all models, which would result in an aggregate-dumping amount of zero.
6.1.2.1.2 With zeroing
Model zeroing comes into play at the second step of the calculations explained above. For each model, whenever the average export price exceeds the average home-market price, the USDOC imputes a zero for those model comparisons.
In the example given in Table 3, only model A is sold at a lower price in the export market than on the domestic market. The dumping amount of model A is 30, and the result generated by applying zeroing is given in column (12). As applied by the USDOC, model zeroing would result in a total dumping amount of 30, which would imply a dumping margin of 3.3%.Footnote 49
It is imperative to note that because of the application of model zeroing, dumping of a single model would result in an affirmative finding of dumping for the product as a whole, even if all other models are not dumped, and even if the negative dumping margins far outweigh the positive margins.
6.1.3 General implication – larger margins
The examples discussed above illustrate the general implication of either simple or model zeroing when combined with any of the methods for calculating dumping margins. Zeroing will almost always inflate the dumping margins and can never lower the margins. Only when all export prices are lower (or higher) than the corresponding comparable home-market prices, will zeroing have no impact – and even in this rare occurrence, zeroing is only neutral (no effect). Thus, in the more typical circumstances when the prices differ in the two markets, zeroing will always raise the calculated margin.
6.1.4 When is zeroing used?
6.1.4.1 Article 5 AD Agreement original investigations
Original antidumping investigations are governed by Article 5 AD Agreement. In the course of such an investigation, a dumping margin is calculated on the basis of the information provided by the exporters during the investigation period. In making this calculation, the administering authorities may decide to resort to simple or model zeroing.
6.1.4.2 Article 9.3.1 AD Agreement reviews
Article 9.3.1 AD Agreement provides that when the amount of the antidumping duty is assessed on a retrospective basis, the determination of the final liability for the payment of antidumping duties shall take place as soon as possible, normally within 12 months, and in no case more than 18 months, after the date on which a request for a final assessment of the amount of the antidumping duty has been made.
While most users of the antidumping instrument impose antidumping duties on a prospective basis, the US uses the retrospective system. In other words, in the US system, the antidumping duty imposed at the end of the original investigation only constitutes an estimate of the future liability. However, at the end of the day, the actual payment of antidumping duties will depend on the calculations made by the USDOC in the course of the annual administrative or duty-assessment reviews.
In the course of such administrative reviews, the USDOC may then decide to resort to simple zeroing when comparing the export transactions with the normal value.
6.1.4.3 Article 11.3 AD Agreement expiry or sunset reviews
Article 11.3 AD Agreement provides that any definitive antidumping duty shall be terminated on a date not later than five years from its imposition, unless the authorities determine, in a review initiated before that date on their own initiative or upon a duly substantiated request made by or on behalf of the domestic industry within a reasonable period of time prior to that date, that the expiry of the duty would be likely to lead to continuation or recurrence of dumping and injury. The duty may remain in force pending the outcome of such a review. Article 11.3 reviews are called expiry or sunset reviews.
The operative test to determine whether the duties should expire is whether the expiry would be likely to lead to a continuation or recurrence of dumping and injury. The application of this test does not necessarily entail a detailed recalculation of the dumping margin. Instead, the authorities may decide to rely upon the dumping margins calculated in the original Article 5 investigation. In such a case, if zeroing, whether model or simple, has been applied in the original investigation, then it effectively percolates down to the Article 11.3 analysis also.
6.1.4.4 Article 11.2 AD Agreement interim reviews
Article 11.2 AD Agreement provides that the authorities shall review the need for the continued imposition of the duty, where warranted, on their own initiative or, provided that a reasonable period of time has elapsed since the imposition of the definitive antidumping duty, upon request by any interested party that submits positive information substantiating the need for a review. Interested parties shall have the right to request the authorities to examine whether the continued imposition of the duty is necessary to offset dumping, whether the injury would be likely to continue or recur if the duty were removed or varied, or both. This type of review is often called interim or changed-circumstances review.
As far as dumping is concerned, the test here is whether the continued imposition of the duty is necessary to offset dumping. The underlying analysis would appear to require a calculation of the dumping margin during the interim-review investigation period, in the process of which the authorities might use either model or simple zeroing.
However, not all jurisdictions recalculate the dumping margins and instead sometimes they base their findings on the calculations made in the original Article 5 investigation. The authorities may then rely upon the dumping margins calculated previously using zeroing techniques.
6.1.4.5 Article 9.5 AD Agreement new-shipper reviews
Article 9.5 AD Agreement provides that the authorities shall promptly carry out a review for the purpose of determining individual margins of dumping for any exporters or producers in the exporting country in question who have not exported the product to the importing Member during the period of investigation, provided that these exporters or producers can show that they are not related to any of the exporters or producers in the exporting country who are subject to the antidumping duties on the product. Such reviews are called new-shipper or newcomer reviews.
In calculating the dumping margin for a new shipper, the authorities may decide to use simple- or model-zeroing techniques.
6.1.5 Prior zeroing decisions of the AB
6.1.5.1 History at the AB
Zeroing has a rather long history of adjudication by the AB. The two cases analyzed in this report represent the fourth and fifth AB dispute in which some aspect of zeroing was adjudicated. As far as we can tell, zeroing is the single most litigated subject in the history of the WTO.
In Table 4, we list the five AB disputes regarding zeroing. As shown, the first three cases had fairly modest scope – each dispute pertained to only original investigations, and each involved a single type of zeroing (either simple or model). Each of the first three cases only involved ‘as applied’ claims, which means that only zeroing as applied in concrete cases was at issue.
Table 4. Typology of zeroing claims in WTO disputes
By contrast, the two AB decisions discussed in this paper are quite broad. While both cases involve original investigations, they also contain claims to include other times during an antidumping case where zeroing can be used. The cases also bring both ‘as applied’ and ‘as such’ claims against the practice of zeroing.
6.1.5.2 Prior AB cases and key findings
6.1.5.2.1 European Communities – Anti-dumping duties on imports of cotton-type bed linen from India, WT/DS141/AB/R of 1 March 2001 (EC–Bed Linen)
In 1999, India challenged the use of model zeroing by the EC in EC–Bed Linen.Footnote 52
The Panel ruled that model zeroing violated Articles 2.4 and 2.4.2 of the AD Agreement and, on appeal, the AB agreed with the findings of the Panel:
By ‘zeroing’ the ‘negative dumping margins’, the European Communities … did not take fully into account the entirety of the prices of some export transactions, namely, those export transactions involving models of cotton-type bed linen where ‘negative dumping margins’ were found. Instead, the European Communities treated those export prices as if they were less than what they were. This, in turn, inflated the result from the calculation of the margin of dumping …. Furthermore, we are also of the view that a comparison between export price and normal value that does not take fully into account the prices of all comparable export transactions – such as the practice of ‘zeroing’ at issue in this dispute – is not a ‘fair comparison’ between export price and normal value, as required by Article 2.4 and by Article 2.4.2.Footnote 53
The AB rejected the argument that model zeroing should be allowed to offset targeted model dumping. In the view of the AB, the exception in Article 2.4.2 only allowed Members to address three kinds of targeted dumping: dumping targeted at certain purchasers, certain regions, or certain time periods. It considered that neither Article 2.4.2 nor any other provision of the AD Agreement referred to dumping targeted at certain models or types of the same product under consideration and noted that ‘had the drafters of the ADA intended to authorize Members to respond to such kind of targeted dumping, they would have done so explicitly in Article 2.4.2, second sentence’.Footnote 54
In rejecting the practice of model zeroing, the AB was emphatic that dumping margins are established for the product concerned as a whole and not just for those exports sold at prices below normal value.Footnote 55 Since the EC had identified the product under consideration as ‘cotton-type bed linen’, the AB found that the EC was bound to treat this product consistently thereafter, in accordance with that definition,Footnote 56 particularly in its establishment of the existence of margins of dumping for the product subject to the investigation.
6.1.5.2.2 United States – Final dumping determination on softwood lumber from Canada, WT/DS264/AB/R of 11 August 2004 (US–Softwood Lumber V)
In US–Softwood Lumber V, Canada had contested model zeroing applied by the USDOC.Footnote 57 While two of the three panelists firmly rejected the practice, one Panel member strongly disagreed with the other two in a dissenting opinion and endorsed the zeroing as applied by the USDOC. On appeal, however, the AB summarily affirmed its findings in EC–Bed Linen and paid no attention to the dissenting opinion:
Zeroing means, in effect, that at least in the case of some export transactions, the export prices are treated as if they were less than what they actually are. Zeroing, therefore, does not take into account the entirety of the prices of some export transactions, namely, the prices of export transactions in those sub-groups in which the weighted average normal value is less than the weighted average export price. Zeroing thus inflates the margin of dumping for the product as a whole.Footnote 58
6.1.5.2.3 United States – Final dumping determination on softwood lumber from Canada, WT/DS264/AB/RW of 15 August 2006 (US–Softwood Lumber V (compliance))
In order to implement the AB ruling in US–Softwood Lumber V, the USDOC recalculated the dumping margins, and, in the new calculations, while using the transaction-to-transaction method, it zeroed the nondumped transactions. This application of simple zeroing was upheld by the Panel in the subsequent compliance case, but the Panel's findings were overturned by the AB:Footnote 59
Turning to the transaction-to-transaction methodology, Article 2.4.2 provides that ‘margins of dumping’ may be established ‘by a comparison of normal value and export prices on a transaction-to-transaction basis’. The reference to ‘export prices’ in the plural suggests that the comparison will generally involve multiple transactions, as was the case in the anti-dumping investigation before us. At the same time, the reference to ‘a comparison’ in the singular suggests an overall calculation exercise involving aggregation of these multiple transactions. The transaction-specific results are mere steps in the comparison process. This tallies with the term ‘basis’ at the end of the sentence, which suggests that these individual transaction comparisons are not the final results of the calculation, but, rather, are inputs for the overall calculation exercise. Thus, the text of Article 2.4.2 implies that the calculation of a margin of dumping using the transaction-to-transaction methodology is a multi-step exercise in which the results of transaction-specific comparisons are inputs that are aggregated in order to establish the margin of dumping of the product under investigation for each exporter or producer. Contrary to the United States' submission, the results of the transaction-specific comparisons are not, in themselves, ‘margins of dumping’.
Furthermore, the reference to ‘export prices’ in the plural, without further qualification, suggests that all of the results of the transaction-specific comparisons should be included in the aggregation for purposes of calculating the margins of dumping. In addition, the ‘export prices’ and ‘normal value’ to which Article 2.4.2 refers are real values, unless conditions allowing an investigating authority to use other values are met. Thus, in our view, zeroing in the transaction-to-transaction methodology does not conform to the requirement of Article 2.4.2 in that it results in the real values of certain export transactions being altered or disregarded.Footnote 60
The AB therefore held that the use of zeroing in the transaction-to-transaction method violated Article 2.4.2 AD Agreement.Footnote 61
6.2 Legal issues and methodologies: did the AB get it right?
6.2.1 Legal basis
6.2.1.1 The zeroing methodology/procedures
In US–Zeroing (EC), the AB ruled that the ‘zeroing methodology’, as it relates to original investigations in which the weighted-average-to-weighted-average comparison method is used to calculate dumping margins, can be challenged ‘as such’.
The AB first discussed whether the zeroing methodology constituted a ‘measure’ within the meaning of Article 3.3 DSU. It considered that the term ‘laws, regulations and administrative procedures’ in Article 18.4 AD Agreement covers all generally applicable rules, norms, and standards adopted by WTO members in connection with the conduct of AD proceedings, whether or not in the form of a written instrument.Footnote 62 As regards challenges to unwritten rules, norms, and standards, it noted that the complaining party would have to clearly establish, through arguments and supporting evidence, (1) that they are attributable to the responding member, (2) their precise content, and (3) that they have general and prospective application. Such evidence could include proof of their systematic application.
The AB considered the evidence before the PanelFootnote 63 sufficient to meet these three requirements and therefore agreed that the zeroing methodology constituted a challengeable measure.
Similarly, in US–Zeroing (Japan), the AB upheld the Panel finding that the ‘zeroing procedures’ (model zeroing in weighted-average-to-weighted-average as well as simple zeroing in transaction-to-transaction comparisons in Article 5 investigations) can be challenged ‘as such’. The AB further considered that the Panel had had sufficient evidence before it to conclude that the zeroing procedures, whether under different comparison methods or at different stages of a proceeding, reflected different manifestations of the same rule or norm, rather than separate rules or norms.
It seems to us that on the basis of the available evidence the Panels and the AB justifiably reached the conclusion that the USDOC's zeroing methodology/procedures amounted to a measure that can be challenged ‘as such’ in WTO dispute-settlement proceedings.
Bearing in mind that most other WTO Members using the AD instrument rely on internal guidelines on issues such as zeroing rather than published or publicly available instruments (such as standard computer programs or antidumping manuals), the US againFootnote 64 became the victim of its transparency.
6.2.1.2 Zeroing in reviews
In US–Zeroing (EC), the AB found that the simple zeroing methodology, ‘as applied’ by the US in 16 administrative duty-assessment reviews was inconsistent with Article 9.3 AD Agreement and Article VI:2 GATT 1994. The AB noted that Article 9.3 AD Agreement referred back to the dumping margin as established under Article 2 AD Agreement and also that it had previously held that the dumping margins are to be established for the product under investigation as such. Additionally, it observed that the antidumping duties assessed under Article 9.3.1 could not exceed the dumping margins calculated for the exporters. The AB considered that the USDOC's simple zeroing practice led to an assessment of duties in excess of the calculated dumping margin because the individual export transactions that were higher-priced than the average normal value were systematically disregarded.
In the same case, the AB overruled the Panel finding that simple zeroing ‘as such’ under the weighted-average-to-transaction method in Article 9.3.1 reviews did not violate Article 9.3 AD Agreement and Article VI:2 GATT, but found itself unable to complete the analysis in the absence of factual Panel findings.
In US–Zeroing (Japan), the AB ruled that simple zeroing in Article 9.3.1 and Article 9.5 reviews ‘as such’ violated Articles 9.3 and 9.5 AD Agreement and Article VI:2 GATT. The AB reversed the Panel finding that simple zeroing in administrative and new-shipper reviews ‘as such’ is not inconsistent with Articles 2.1, 9.1, and 9.2 AD Agreement and Article VI:1 GATT, but did not find it necessary to complete the analysis.
The AB further found that the simple zeroing as applied in the 11 administrative reviews violated Articles 2.4 and 9.3 AD Agreement and Article VI:2 GATT 1994. As regards the Articles 2.1 and 9.1 claims, it reversed the relevant Panel findings but did not find it necessary to complete the analysis.
Additionally, in the same case, the AB also ruled that the US had infringed Article 11.3 AD Agreement by relying, in two expiry reviews, on the dumping margins calculated in administrative reviews using zeroing.
As far as the AB jurisprudence with regard to zeroing in reviews is concerned, we make three observations:
First, it seems to us that the AB very easily dismisses the importance of the phrase, ‘during the investigation phase’, mentioned in Article 2.4.2 AD Agreement. The AD Agreement appears to clearly distinguish between Article 5 original investigations on the one hand and reviews on the other hand, see notably Article 18.3 AD Agreement. While reviews typically also involve ‘investigations’, the AD Agreement does not use the term in this context. The term ‘investigation phase’ in Article 2.4.2 AD Agreement is also clearly different from the term ‘period of investigation’ in other parts of Article 2 AD Agreement. Thus, it should have a different meaning. Therefore, it would appear to us that the phrase must operate as a temporal limitation on the scope of Article 2.4.2 by excluding reviews from the purview of its applicability (although this does not automatically mean that zeroing is allowed in reviews). Indeed, it seems relatively clear, for example, that if the authorities in the course of an expiry or interim review rely on illegally calculated dumping margins of the original investigation, such reliance automatically makes the review findings illegal as well. If anything, this is a simple application of the fruit-of-the-poisoned-tree concept. However, it would entail that zeroing in reviews cannot be considered to violate Article 2.4.2 AD Agreement, although it would still violate Article 9.3 AD Agreement.
Second, as regards administrative reviews in a retrospective system, the AB very readily dismisses the US arguments that a prohibition of zeroing in retrospective systems favors prospective systems. The AB notes that any form of zeroing in a prospective duty-collection system would allow the importer to claim a refund.Footnote 65 However, this ignores the reality that in most prospective systems, refund applications and refunds granted are rare (and oftenFootnote 66 not even published). Furthermore, in prospective systems, duties are sometimes imposed in the form of a variable duty or on the basis of prospective normal values. In such cases, duties are payable automatically by the importers whenever they import at a price below the benchmark price as established in the definitive-duty determination. Because the prices of export transactions made above the benchmark price will not offset the prices of transactions made at prices below the benchmark price, zeroing effectively takes place, too. Yet, such zeroing would appear to be not illegal per se because, per the AB, the importer could in theory qualify for a refund.
Third, we would note that the exporter-oriented approach, which the AB considers the correct one, will not necessarily lead to better results for individual importers than the importer-based US approach. Thus, under the US approach, an importer that imports at nondumped prices will not have to pay any duties. However, under an exporter-based approach, the authorities will calculate a weighted-average dumping margin. Suppose that in the original investigation the margin was 10%, but in the duty-assessment review it turns out to be 7%, then all importers will have to pay 7%, irrespective of their actual import-price levels.
Whether this concern manifests itself in practice is unclear. From our review of US administrative reviews, it appears that, most typically, reviews with duty reductions involve a single importer buying from a single exporter rather than a large set of importers purchasing from one or more suppliers. This is likely because of the rather large risk importers assume under the retrospective system; it is the importers who are responsible for the bond and perhaps ex post higher duty, but it is the exporters' pricing behavior that triggers the duty reassessment. As a result, both sides of the transaction – the importer and exporter – have to commit to managing the duty. Apparently, this commitment is deemed too costly (or the review process too uncertain) when there are many importers.
6.2.1.3 The violation of Article 2.4 AD Agreement
In US–Zeroing (Japan), the AB ruled that simple zeroing in Article 9.3.1 and Article 9.5 reviews ‘as such’ violated Article 2.4 AD Agreement. The AB further found that the simple zeroing ‘as applied’ in the 11 administrative reviews violated Articles 2.4 and 9.3 AD Agreement and Article VI:2 GATT 1994. Lastly, the AB also found that simple zeroing under the transaction-to-transaction method in original investigations ‘as such’ violated the fair-comparison requirement of Article 2.4 and overruled the Panel finding that Article 2.4 could not be interpreted in such a way as to make the more specific Article 2.4.2 provisions inoperative.
We believe that the AB very easily finds that zeroing violates the fair-comparison requirement of Article 2.4 AD Agreement. The fair-comparison requirement, as used in Article 2.4, encompasses not only the actual comparison between normal value and export price, but also the entire netting-back process on both sides. It is a guiding principle, but it also is inherently vague. 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.
Moreover, the finding that zeroing violates Article 2.4 necessarily implies that zeroing in whatever form is also prohibited under the exceptional method. If so, then depending on exactly how the targeted and nontargeted dumping is identified and how the administering authorities compute the comparison weighted average, it may well be the case that the exceptional method has no meaning, as the Panel correctly points out.
Furthermore, the priority that the AB assigns to Article 2.4 over the more specific provisions of Article 2.4.2 may be contrasted with the jurisprudence concerning the relationship between the more general Article 2.2 and the more specific Article 2.2.2 AD Agreement, where Panels have repeatedly held that Article 2.2 does not impose an overarching reasonableness requirement and that the specific Article 2.2.2 methods for determining SGA and profit in constructed normal values are by definition reasonably within the meaning of Article 2.2.Footnote 67
In summary, it would appear to us that zeroing is legally allowed under the exceptional method of Article 2.4.2, but that the AB's ruling that zeroing violates the fair-comparison requirement effectively prohibits it.
6.3 Economic issues and methodologies
6.3.1 Statistical bias
In prior commentaries on the earlier AB decisions on zeroing, Janow and Staiger and Bown and Sykes wrestle with understanding the economic rationale for antidumping, let alone zeroing:
Does each ‘dumped’ sale of a product create a similarly unacceptable risk of economic harm that deserves sanction, regardless of the pricing behavior of the firm in other transactions? To answer this question, one must embrace a theory of why dumping is worrisome in the first instance. But modern economic learning has difficulty explaining why dumping should be subject to sanction at all.Footnote 68
… the goal of Article VI GATT as it relates to anti-dumping duties (Article VI GATT provides as well for countervailing duties) is, if not to discourage or prevent outright the practice of dumping in international trade, then at the very least to provide governments with the ability to shield their producers from the effects of dumping with extraordinary tariff responses.
The tariff responses to dumping provided for in Article VI GATT are extraordinary not so much because they permit governments to raise tariffs above their bound levels in the face of import-induced injury – there are a variety of other ‘safeguard’ provisions that might be utilized by a WTO member government to achieve this – but because they allow for discriminatory tariffs to be imposed and do not provide for the government of the country from which the dumped exports originate to seek compensation.
From a standard economic perspective, it is very difficult to make sense of the goal suggested by a reading of Article VI GATT. Unless dumping is truly predatory, which in practice appears rarely to be the case, there is no standard efficiency rationale for the position that dumped imports should be treated any differently by a government than imports that are not dumped. Dumped or not, a given volume of imports will have the same impact on prices and incomes in the domestic economy once it crosses the border.Footnote 69
At the end of the day, neither Bown and Sykes nor Janow and Staiger were able to produce an economic rationale for antidumping as currently implemented (i.e., antidumping is not tied to an economic theory of predation). Despite the lack of a guiding economic theory for the law, both pairs of authors conclude that while they largely agreed with the AB decision – that zeroing is inconsistent with AD Agreement – they disagreed with the legal justifications that the AB used in coming to this conclusion. And this is essentially our conclusion also.
At a fundamental level, the economic problem of zeroing is that it conflicts with well-established econometric methods for producing unbiased estimates. Without a redrafting of the language governing the scope of the analysis, the claim that zeroing produces an unbiased comparison seems dubious and perhaps downright disingenuous. Econometricians have spent decades thinking carefully about bias and consistency in data analysis. They obsessively worry about the accuracy of inference because of outliers. At some level, the justification for zeroing is the need to properly account for outliers. Advocates of zeroing argue that (1) it is the low-price transactions that really cause the injury, and (2) in order to accurately capture the impact of the low-price transactions one has to discard the high-price transactions. But such an approach violates basic econometric principles.
By analogy, suppose one is trying to determine if women earn lower wages than comparable men. To answer this question, one might collect data on a large set of men and women, and then match them up according to known characteristics (age, type of job, tenure, highest education level, etc.) Then, after adjusting for these factors to create comparable men and women, one might compare the wages for each matched pair – or more likely compare the weighted-average difference in wages. Whatever the precise statistical approach taken, ordinary least squares, quantile regression, Tobit, etc., the econometrician would include all observations.
In the context of this example, zeroing would imply that the econometrician discard all observations where a woman received a higher wage than her comparable man; then, after dropping these observations, the econometrician would rerun the statistical analysis with only matched pairs that involved the woman making the same or less wage than her comparable man. It would come as no surprise that such an approach would indeed show that women make less than men. Whether one normalized the implied wage differences by the total sample size or by only those in the reduced sample would hardly matter to the core issue – the validity of the inference. Further, even including all matched pairs in the normalization step (whereby the econometrician computes the percentage wage difference) would not justify dropping a set of observations because the guiding principle for dropping the matched pair is intimately related to the question that the econometrician is trying to answer.
Any person doing such a ‘data mining’ procedure would be subject to severe professional ridicule, and the results would be ignored as biased. The same conclusion must apply to the method of zeroing. A method that can only serve to raise margins and can never lower a margin seemingly has a clear motive – to generate margins so duties can be imposed on exporters. Zeroing can render a large number of actual export transaction prices moot for they are treated as if they were made at lower prices than they actually were. Given that the objective of the accounting exercise is to ascertain whether the exporter has sold at an unfair low price, zeroing has the effect of lowering the exporter's measured price.
The notion that, without zeroing, authorities cannot adequately assess the impact of certain low-priced transactions is false. Econometricians have developed a large body of work precisely to measure accurately the impact of outliers on the statistic of interest. None of these methods truncate or ‘throw away’ just positive outliers.
6.3.1.1 Can one instance of dumping be demonstrative?
Above we argued that Article 2.4.2's requirement that the margins be calculated for the product under investigation as a whole is analogous to the standard econometric requirement that all observations be used when performing statistical tests. Zeroing leads to biased estimates of the margin based on the entirety of the product.
The question remains whether it is possible that a single instance (or handful of instances) of dumping could be sufficiently injurious to justify zeroing. The argument is that unless authorities focus on just the set of transactions at low prices, they will miss the impact. If this hypothesis is correct, then this might imply that zeroing is necessary for authorities to identify and appropriately sanction injurious dumping. For instance, suppose the domestic industry's ‘normal’ profit margin is very small. One can imagine that in this case even a few transactions with low prices (e.g., sales below cost), for even a very short period of time, might lead the domestic firm (or industry) to go out of business. Or at the minimum, it could take the industry a very long time to earn enough profits to offset the injury incurred because of the dumping.
We do not find this argument a compelling justification for zeroing. To begin with, the ADA does have a provision to account for such a circumstance – the exceptional method. In our hypothetical example, a few transactions may be inordinately significant to the margin calculation. But, this simply provides justification for the exceptional method rather than zeroing. Secondly, the argument is primarily about the sensitivity of injury to small margins, rather than a need to increase the size of the margin. Again, the notion that authorities need flexibility in assessing injury is unrelated to zeroing. Finally, the ADA provides for investigations to be terminated should the dumping margin be de minimis – less than 2%. Whether the de minimis standard is too high, again does not mean zeroing is justified.
6.3.2 Methods for calculating margins
One argument made justifying zeroing is that without zeroing the weighted-average-to-weighted-average method and the weighted-average-to-transaction methods will yield the same margins. To see this, let p i denote the home-market price, p i* denote the export price, and q i the volume exported. Also assume that we have a total of N comparable transactions.
Now if we assume that the weighted-average part of the exceptional method is based on all transactions, then we can show thatFootnote 70
Under this approach toward the exceptional method, authorities would first calculate the weighted-average home-market price. This is the term in parentheses in the above equation. Second, this single price will be compared with each individual transaction to compute a dumping amount (the bracketed terms). Then, the various margins would be weighted by volume. As shown, under this approach the exceptional method produces the same margin as the preferred WW method.
What does this mean? Under at least some formulation of the exceptional method, it, too, yields the same margin as the preferred weighted-average-to-weighted-average method. Because of this possible equivalency, it can be argued that zeroing must be permissible under Article 2.4.2, at least under the exceptional method. We note, however, that it is not the case that the exceptional method (without zeroing) will always produce the same margin as the preferred methods. The equivalency is not robust to other methods of computing the dumping amount under the exceptional method. The exceptional method can produce different dumping amounts than the preferred methods even without invoking zeroing. We demonstrate this possibility below.
7. Concluding comments and thoughts on what's next for zeroing
7.1 Conclusions
The AB has come down hard on zeroing, virtually prohibiting the practice no matter its form (model or simple), the comparison method (weighted-average-to-weighted-average, transaction-to-transaction, or weighted-average-to-transaction), or the stage of the proceeding (original investigation, administrative review, expiry review, new-shipper review, or interim review).
The AB has done so by interpreting general concepts (the calculation of one dumping margin per exporter for the product concerned; the obligation not to levy an antidumping duty greater than the dumping margin; the fair-comparison requirement) and declaring their applicability throughout the proceeding.
In this process, the AB has arguably rendered certain provisions of the AD Agreement inutile, notably the term ‘during the investigation phase’ in Article 2.4.2 and possibly the entire exceptional comparison method under Article 2.4.2.Footnote 71
It seems to us that where simple zeroing is employed, model zeroing is not an issue because, in the calculation of the dumping margin, there is no intermediate step of calculating dumping amounts or margins per model. In effect, then the total dumping amounts (after the exclusion of higher-prices export sales) are divided by the total denominator to calculate the margin as a percentage. This observation, if conceptually correct, is important because it leads us to conclude, differently from Bown and Sykes,Footnote 72 that model zeroing was a new method developed by the EC and the US, because they understood that after the entry into force of the AD Agreement, they could no longer resort to simple zeroing under the preferred comparison methods of Article 2.4.2 AD Agreement.Footnote 73 Most other countries understood that they had limited the use of zeroing to the exceptional method and therefore did not hesitate to challenge the newly developed model-zeroing concept.
7.2 The way forward under the exceptional method
7.2.1 How to define/identify ‘targeted’ dumping
It is difficult to conceive of how the logic invoked by the AB in the context of zeroing under the other two methods would cease to apply for the exceptional-comparison method under which the individual export transactions are compared to an average normal value. If disregarding certain export transactions because they are higher-priced than the domestic transactions with which they are compared is forbidden under the transaction-to-transaction approach, it would seem inconsistent to ignore the identical logic and not reach the identical conclusion with respect to the exceptional method, which, too, considers individual export transactions.
It is possible, as the AB suggests in US–Zeroing (Japan), that there may never be the need to invoke the exception provided in Article 2.4.2. The condition of a ‘pattern of export prices which differ significantly among different purchasers, regions or time periods’ might never materialize, or, if that pattern exists, it still might be taken into account by one of the two preferred methodologies set out in the first sentence of Article 2.4.2. Certainly, that would seem to be possible where there are price differences between time periods. Transaction-to-transaction comparisons seem particularly well-suited to overcoming any potential skewing of the overall margin caused by significant price fluctuations over the period. Likewise, there is nothing wrong with creating multiple averaging periods in both markets for determining the margin of dumping as long as it results in a fair comparison.Footnote 74
Addressing targeted dumping with reference to particular purchasers or regions would appear to be more complicated under the preferred methods, since there is almost certainly no appropriate corresponding subgroup in the domestic market. In fact, these very circumstances are most likely to warrant the use of the exceptional method, but without zeroing, it is argued, the results would be the same as those produced under the weighted-average-to-weighted-average method.
However, the AB seems to have an alternative suggestion. Although the specific question of zeroing under the exception was not before the AB in US–Zeroing (Japan), the AB felt compelled to address the mathematical-equivalence argument. To support its own implication that the exception without zeroing could yield a unique result, the AB noted that:
The emphasis in the second sentence of Article 2.4.2 is on a ‘pattern’, namely a ‘pattern of export prices which differs significantly among different purchasers, regions or time periods’. The prices of transactions that fall within this pattern must be found to differ significantly from other export prices. We therefore read the phrase ‘individual export transactions’ in that sentence as referring to the transactions that fall within the relevant pricing pattern. This universe of export transactions would necessarily be more limited than the universe of export transactions to which the symmetrical comparison methodologies in the first sentence of Article 2.4.2 would apply. In order to unmask targeted dumping, an investigating authority may limit the application of the W-T comparison methodology to the prices of export transactions falling within the relevant pattern.Footnote 75
The AB seems to be suggesting a hybrid methodology, whereby sales not exhibiting a pattern of significant price differences would be compared using one of the two preferred methods, and those exhibiting significant differences would be subject to the exceptional method. A prohibition of zeroing under this manifestation of the exception would not necessarily produce a result mathematically equivalent to that stemming from use of one of the preferred methodologies.
One possible approach will be for authorities to look at the overall distribution of prices and identify those purchasers, regions, or time periods with unusually low prices. An obvious metric would be to look at transactions that are two or more standard deviations from the overall sample weighted average, although it appears that the USDOC's first effort to utilize the exceptional method deems prices that are one standard deviation below the mean to be ‘exceptional’.
The next step is the determination whether there is any sort of pattern to the outliers. Finally, if the authorities determine a pattern exists, they will need to provide an explanation why such differences cannot be taken into account appropriately by one of the preferred methods.
7.2.2 How to calculate a margin
Let's return to the same transaction pricing data used in our previous discussion of zeroing. Now let's assume that the product was sold to four customers, W, X, Y, and Z (Table 5). Customer W purchased on 2 and 4 September, customer X purchased on 8 and 10 September, customer Y purchased on 12 and 16 September, and customer Z purchased on 18, 20, and 24 September. As we discussed earlier, the weighted-average price for home-market transactions is 100.
Table 5. The exceptional method
The standard deviation of the export prices given in Table 5 is 16 (column (2)). Using the one standard-deviation approach tentatively adopted by the USDOC, any transaction price less than 84 (100–16) would be considered unusually low. In our example, both of customer W's purchases would be deemed to have been targeted.
How might the USDOC implement the exceptional method without imposing zeroing? We offer three possibilities.
Method 1 – First, we assume the authorities use the weighted-average-to-transaction method for the targeted customer and the weighted-average-to-weighted-average method for all other customers. The home-market price for both the targeted customers and for all others is based on all comparable transactions. As shown in column (14), customer W would have a dumping amount of 25 for each transaction (100–75). For the other customers, the authorities could calculate the weighted-average export price based just on their set of prices which is 107.Footnote 76 This would lead to a dumping amount for each transaction of −7. Overall, without zeroing the dumping margin is zero.Footnote 77 Thus, in this method if zeroing is not permitted, the exceptional method produces the exact same result as the preferred methods.
Method 2 – Here we again imagine the authorities use the weighted-average-to-transaction method for the targeted customer and the weighted-average-to-weighted-average for all other customers. We again assume the home-market price for both the targeted customers and for all others are based on all comparable transactions. As shown in column (15), customer W would have a dumping amount of 25 for each transaction. For the other customers, the authorities could calculate the weighted-average export price based on export prices, which is 100. This would lead to a dumping amount for each transaction of 0. Overall, without zeroing the dumping margin is 5.6% (50/900).
Method 3 – Finally, we imagine the authorities use the weighted-average-to-transaction method for the targeted customer and the transaction-to-transaction method for all other customers. As with the other two hypothetical methods, we again assume the home-market price for both the targeted customers and for all others are based on all comparable transactions. In this case, customer W would have a dumping amount of 25 for each transaction, yielding a positive dumping amount of 50. For the other customers, the authorities would compute the dumping margin on a transaction-to-transaction basis. This would lead to a negative dumping amount of −35 as indicated in column (16). Overall, after the authorities aggregate all the dumping amounts, they would find an overall dumping amount of 15 (50–35) and a dumping margin of 1.7% (15/900).
As these examples demonstrate, the exceptional method does not require zeroing to produce margins in response to ‘unusually’ low prices. Exactly how countries implement the exceptional method and whether these approaches are WTO-consistent await future AB disputes.
