WTO | WTO analytical index: Guide to WTO Law and Practice

español français

home > resources > publications > wto analytical index > table of contents > anti-dumping agreement

WTO ANALYTICAL INDEX: ANTI-DUMPING AGREEMENT

Agreement on Implementation of Article VI of the General Agreement on Tariffs and Trade 1994 (Anti-Dumping Agreement)

The texts reproduced here do not have the legal standing of the original documents which are entrusted and kept at the WTO Secretariat in Geneva.

> Article 1
> Article 2
> Article 3
> Article 4
> Article 5
> Article 6
> Article 7
> Article 8
> Article 9
> Article 10
> Article 11
> Article 12
> Article 13
> Article 14
> Article 15
> Article 16
> Article 17
> Article 18
> Annex I
> Annex II
> Relationship with other WTO Agreements
> Declaration on Dispute Settlement Pursuant to the Agreement on Implementation of Article VI of the General Agreement on Tariffs and Trade 1994 or Part V of the Agreement on Subsidies and Countervailing Measures
> Decision on Review of Article 17.6 of the Agreement on Implementation of Article VI of the General Agreement on Tariffs and Trade 1994
> Decision on Anti-Circumvention

> Analytical Index main page

Part I

I. Article 1 back to top

A. Text of Article 1

Members hereby agree as follows:

Article 1(1): Principles

An anti-dumping measure shall be applied only under the circumstances provided for in Article VI of GATT 1994 and pursuant to investigations initiated(1) and conducted in accordance with the provisions of this Agreement. The following provisions govern the application of Article VI of GATT 1994 in so far as action is taken under anti-dumping legislation or regulations.

(footnote original) 1 The term “initiated” as used in this Agreement means the procedural action by which a Member formally commences an investigation as provided in Article 5.

B. Interpretation and Application of Article 1

1. General

(a) “anti-dumping measure”

1. The Appellate Body in US — 1916 Act rejected the argument that, based on the history of Article 1, “the phrase ‘anti-dumping measure’ refers only to definitive anti-dumping duties, price undertakings and provisional measures.”(2) The Appellate Body stated that “the ordinary meaning of the phrase ‘antidumping measure’ seems to encompass all measures taken against dumping. We do not see in the words ‘an anti-dumping measure’ any explicit limitation to particular types of measures.”(3)

(b) “initiated and conducted in accordance with the provisions of this Agreement”

2. Regarding a claim raised under Article 1, the Panel on US — 1916 Act (EC) noted that “if we find a violation of other provisions of the Anti-Dumping Agreement, it will be demonstrated that the anti-dumping investigation … is not ‘initiated and conducted in accordance with the provisions of this Agreement’ and a breach of Article 1 will be established.” (4)

3. The Panel on EC — Tube or Pipe Fittings rejected the assertion that in case of a devaluation in the fourth quarter of the period of investigation, Article 1 of the Anti-Dumping Agreement and Article VI of the GATT 1994 require the investigating authority to base its determination only on the period following the devaluation to examine whether there was present dumping causing injury. The Panel stated that “Article 1 of the Anti-Dumping Agreement does not require an investigating authority to re-assess its own determination made on the basis of an examination of data pertaining to the IP prior to the imposition of an anti-dumping measure in the light of an event that occurred during the IP”.(5)

4. In EC — Bed Linen, the Panel touched on the relationship between Articles 1 and 15 in interpreting Article 15. See paragraph 585 below.

5. In Guatemala — Cement II, the Panel found that the subject anti-dumping duty order of Guatemala was inconsistent with Articles 3, 5, 6, 7, 12, and paragraph 2 of Annex I of the Anti-Dumping Agreement. The Panel then opined that Mexico’s claims under other articles of the Anti-Dumping Agreement, including Article 1, were “dependent claims, in the sense that they depend entirely on findings that Guatemala has violated other provisions of the AD Agreement. There would be no basis to Mexico’s claims under Articles 1, 9 and 18 of the AD Agreement, and Article VI of GATT 1994, if Guatemala were not found to have violated other provisions of the AD Agreement.”(6) In light of this dependent nature of Mexico’s claim, the Panel considered it not necessary to address these claims.

6. In US — Stainless Steel, addressing Korea’s claim that “because certain provisions of the AD Agreement have been violated, Article VI of the GATT 1994 and Article 1 of the AD Agreement are consequently violated”(7), the Panel also stated: “[b]ecause of their dependent nature, we can perceive of no useful purpose that would be served by ruling on these claims. Accordingly, we do not consider it necessary to address them.”(8)

7. The relationship between Article 1 and other provisions of the Anti-Dumping Agreement was discussed in Guatemala — Cement II and US — Stainless Steel. See paragraphs 5–6 above.

II. Article 2 back to top

A. Text of Article 2

Article 2: Determination of Dumping

2.1 For the purpose of this Agreement, a product is to be considered as being dumped, i.e. introduced into the commerce of another country at less than its normal value, if 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.

2.2 When there are no sales of the like product in the ordinary course of trade in the domestic market of the exporting country or when, because of the particular market situation or the low volume of the sales in the domestic market of the exporting country(2), such sales do not permit a proper comparison, the margin of dumping shall be determined by comparison with a comparable price of the like product when exported to an appropriate third country, provided that this price is representative, or with the cost of production in the country of origin plus a reasonable amount for administrative, selling and general costs and for profits.

(footnote original) 2 Sales of the like product destined for consumption in the domestic market of the exporting country shall normally be considered a sufficient quantity for the determination of the normal value if such sales constitute 5 per cent or more of the sales of the product under consideration to the importing Member, provided that a lower ratio should be acceptable where the evidence demonstrates that domestic sales at such lower ratio are nonetheless of sufficient magnitude to provide for a proper comparison.

2.2.1 Sales of the like product in the domestic market of the exporting country or sales to a third country at prices below per unit (fixed and variable) costs of production plus administrative, selling and general costs may be treated as not being in the ordinary course of trade by reason of price and may be disregarded in determining normal value only if the authorities(3) determine that such sales are made within an extended period of time(4) in substantial quantities(5) and are at prices which do not provide for the recovery of all costs within a reasonable period of time. If prices which are below per unit costs at the time of sale are above weighted average per unit costs for the period of investigation, such prices shall be considered to provide for recovery of costs within a reasonable period of time.

(footnote original) 3 When in this Agreement the term “authorities” is used, it shall be interpreted as meaning authorities at an appropriate senior level.

(footnote original) 4 The extended period of time should normally be one year but shall in no case be less than six months.

(footnote original) 5 Sales below per unit costs are made in substantial quantities when the authorities establish that the weighted average selling price of the transactions under consideration for the determination of the normal value is below the weighted average per unit costs, or that the volume of sales below per unit costs represents not less than 20 per cent of the volume sold in transactions under consideration for the determination of the normal value.

2.2.1.1 For the purpose of paragraph 2, costs shall normally be calculated on the basis of records kept by the exporter or producer under investigation, provided that such records are in accordance with the generally accepted accounting principles of the exporting country and reasonably reflect the costs associated with the production and sale of the product under consideration. Authorities shall consider all available evidence on the proper allocation of costs, including that which is made available by the exporter or producer in the course of the investigation provided that such allocations have been historically utilized by the exporter or producer, in particular in relation to establishing appropriate amortization and depreciation periods and allowances for capital expenditures and other development costs. Unless already reflected in the cost allocations under this sub-paragraph, costs shall be adjusted appropriately for those non-recurring items of cost which benefit future and/ or current production, or for circumstances in which costs during the period of investigation are affected by start-up operations.(6)

(footnote original) 6 The adjustment made for start-up operations shall reflect the costs at the end of the start-up period or, if that period extends beyond the period of investigation, the most recent costs which can reasonably be taken into account by the authorities during the investigation.

2.2.2 For the purpose of paragraph 2, the amounts for administrative, selling and general costs and for profits shall be based on actual data pertaining to production and sales in the ordinary course of trade of the like product by the exporter or producer under investigation. When such amounts cannot be determined on this basis, the amounts may be determined on the basis of:

(i) the actual amounts incurred and realized by the exporter or producer in question in respect of production and sales in the domestic market of the country of origin of the same general category of products;

(ii) the weighted average of the actual amounts incurred and realized by other exporters or producers subject to investigation in respect of production and sales of the like product in the domestic market of the country of origin;

(iii) any other reasonable method, provided that the amount for profit so established shall not exceed the profit normally realized by other exporters or producers on sales of products of the same general category in the domestic market of the country of origin.

2.3 In cases where there is no export price or where it appears to the authorities concerned that the export price is unreliable because of association or a compensatory arrangement between the exporter and the importer or a third party, the export price may be constructed on the basis of the price at which the imported products are first resold to an independent buyer, or if the products are not resold to an independent buyer, or not resold in the condition as imported, on such reasonable basis as the authorities may determine.

2.4 A fair comparison shall be made between the export price and the normal value. This comparison shall be made at the same level of trade, normally at the exfactory level, and in respect of sales made at as nearly as possible the same time. Due allowance shall be made in each case, on its merits, for differences which affect price comparability, including differences in conditions and terms of sale, taxation, levels of trade, quantities, physical characteristics, and any other differences which are also demonstrated to affect price comparability.(7) In the cases referred to in paragraph 3, allowances for costs, including duties and taxes, incurred between importation and resale, and for profits accruing, should also be made. If in these cases price comparability has been affected, the authorities shall establish the normal value at a level of trade equivalent to the level of trade of the constructed export price, or shall make due allowance as warranted under this paragraph. The authorities shall indicate to the parties in question what information is necessary to ensure a fair comparison and shall not impose an unreasonable burden of proof on those parties.

(footnote original) 7 It is understood that some of the above factors may overlap, and authorities shall ensure that they do not duplicate adjustments that have been already made under this provision.

2.4.1 When the comparison under paragraph 4 requires a conversion of currencies, such conversion should be made using the rate of exchange on the date of sale(8), provided that when a sale of foreign currency on forward markets is directly linked to the export sale involved, the rate of exchange in the forward sale shall be used. Fluctuations in exchange rates shall be ignored and in an investigation the authorities shall allow exporters at least 60 days to have adjusted their export prices to reflect sustained movements in exchange rates during the period of investigation.

(footnote original) 8 Normally, the date of sale would be the date of contract, purchase order, order confirmation, or invoice, whichever establishes the material terms of sale.

2.4.2 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.

2.5 In the case where products are not imported directly from the country of origin but are exported to the importing Member from an intermediate country, the price at which the products are sold from the country of export to the importing Member shall normally be compared with the comparable price in the country of export. However, comparison may be made with the price in the country of origin, if, for example, the products are merely transshipped through the country of export, or such products are not produced in the country of export, or there is no comparable price for them in the country of export.

2.6 Throughout this Agreement the term “like product” (“produit similaire”) shall be interpreted to mean a product which is identical, i.e. alike in all respects to the product under consideration, or in the absence of such a product, another product which, although not alike in all respects, has characteristics closely resembling those of the product under consideration.

2.7 This Article is without prejudice to the second Supplementary Provision to paragraph 1 of Article VI in Annex I to GATT 1994.

B. Interpretation and Application of Article 2

1. General

(a) Period of data collection

(i) Recommendation by the Committee on Anti-Dumping Practices

8. At its meeting of 4–5 May 2000, regarding appropriate periods of data collection, the Committee on Anti-Dumping Practices recommended with respect to original investigations to determine the existence of dumping and consequent injury:

“1. As a general rule:

(a) the period of data collection for dumping investigations normally should be twelve months, and in any case no less than six months, ending as close to the date of initiation as is practicable;

(b) the period of data collection for investigating sales below cost, and the period of data collection for dumping investigations, normally should coincide in a particular investigation;

(c) the period of data collection for injury investigations normally should be at least three years, unless a party from whom data is being gathered has existed for a lesser period, and should include the entirety of the period of data collection for the dumping investigation;

(d) In all cases the investigating authorities should set and make known in advance to interested parties the periods of time covered by the data collection, and may also set dates certain for completing collection and/or submission of data. If such dates are set, they should be made known to interested parties.

2. In establishing the specific periods of data collection in a particular investigation, investigating authorities may, if possible, consider practices of firms from which data will be sought concerning financial reporting and the effect this may have on the availability of accounting data. Other factors that may be considered include the characteristics of the product in question, including seasonality and cyclicality, and the existence of special order or customized sales.

3. In order to increase transparency of proceedings, investigating authorities should include in public notices or in the separate reports provided pursuant to Article 12.2 of the Agreement, an explanation of the reason for the selection of a particular period for data collection if it differs from that provided for in: paragraph 1 of this recommendation, national legislation, regulation, or established national guidelines.”(9)

(ii) The role of the investigation period

9. The Appellate Body on EC — Tube or Pipe Fittings rejected an argument made by Brazil that the investigating authority was obliged to base its export price determination on data relating to only that part of the period of investigation that followed an important devaluation of the Brazilian currency. According to the Appellate Body “certain anomalous results would flow from Brazil’s assertion that when a major change, such as in this case a steep and lasting devaluation, occurs at a late stage of the POI, the dumping determination should be confined to and based on the data following that major change. If such a change were to take place at the very end of the POI, Brazil’s approach would imply that the determination would have to be based on the data of a very short period.”(10) The Appellate Body reached the following conclusion with regard to the role of the period of investigation:

“Permitting such discretionary selection of data from a period of time within the POI would defeat the objectives underlying investigating authorities’ reliance on a POI for the purposes of a dumping determination. As the Panel correctly noted, the POI ‘form[s] the basis for an objective and unbiased determination by the investigating authority.’ Like the Panel and the parties to this dispute, we understand a POI to provide data collected over a sustained period of time, which period can allow the investigating authority to make a dumping determination that is less likely to be subject to market fluctuations or other vagaries that may distort a proper evaluation. We agree with the Panel that the standardized reliance on a POI, although not fixed in duration by the Anti-Dumping Agreement, assures the investigating authority and exporters of ‘a consistent and reasonable methodology for determining present dumping’, which anti-dumping duties are intended to offset. In contrast to this consistency and reliability, Brazil’s approach would introduce a significant level of subjectivity on the part of the investigating authority to determine when data from a subset of the POI may be a reliable indicator of an exporter’s future pricing behaviour. As the European Communities points out, the ‘broad judgmental role’ accorded investigating authorities by Brazil’s approach is not consistent with the detailed nature of the rules and obligations of the Anti-Dumping Agreement governing various aspects of the dumping determination.”(11)

10. The Appellate Body in EC — Tube or Pipe Fittings further considered that “the Anti-Dumping Agreement takes into account the possibility of such major changes occurring at a late stage of the POI, or even after the POI, not by allowing investigating authorities to pick and choose a subset of data or sub-periods of a POI according to their subjective considerations, but by review mechanisms.”(12)

(b) Relationship with other paragraphs of Article 2

11. In US — Stainless Steel, the Panel found the United States treatment of unpaid export sales as direct selling costs to be inconsistent with Article 2.4. In the context of this finding, the Panel explained the relationship between Articles 2.1, 2.3 and 2.4, as follows:

“In our view, both Article 2.3 and Article 2.4 play an important role in respect of the construction of export prices. When determining whether dumping exists, Article 2.1 usually requires a comparison of the export price with the comparable price, in the ordinary course of trade, for the like product when destined for consumption in the exporting country. Article 2.3, however, authorizes a Member to construct the export price where, inter alia, the actual export price is unreliable because of association between the exporter and the importer. As discussed in section VI.C.2.(b)(i), it was pursuant to this authorization that the DOC disregarded the export price charged by POSCO to its affiliated importer POSAM in these investigations and instead constructed the export price.

Further, Article 2.3 specifies that the export price may be constructed on the basis of the price at which the imported products are first resold to an independent buyer. It is clear from this language that, while the price charged to the first independent buyer is a starting-point for the construction of an export price, it is not itself the constructed export price. Nor does Article 2.3 itself contain any guidance regarding the methodology to be employed in order to construct the export price. Rather, the only rules governing the methodology for construction of an export price are set forth in Article 2.4 of the AD Agreement, which provides that, ‘[i]n the cases referred to in paragraph 3, allowances for costs, including duties and taxes, incurred between importation and resale, and for profits accruing, should also be made.’ Although the United States repeatedly refers to these allowances as ‘Article 2.3 adjustments’, the provision governing these allowances is found in Article 2.4 and it is therefore evident to us that a claim regarding the appropriateness of allowances made to construct an export price may be made pursuant to that Article.(13)”(14)

2. Article 2.1

(a) Conditions on sales transactions for the calculation of normal value

12. In US — Hot-Rolled Steel, the Appellate Body considered that “[t]he text of Article 2.1 expressly imposes four conditions on sales transactions in order that they may be used to calculate normal value: first, the sale must be ‘in the ordinary course of trade’; second, it must be of the ‘like product’; third, the product must be ‘destined for consumption in the exporting country’; and, fourth, the price must be ‘comparable’”.(15)

(i) Use of downstream sales

13. In US — Hot-Rolled Steel, the United States authorities, in calculating the normal value, discarded certain sales by exporters to their affiliates because these sales were not “in the ordinary course of trade”. The authorities replaced the discarded sales with downstream sales of the product, transacted between the affiliate and the first independent buyer, which had been made “in the ordinary course of trade”. Japan objected to the use of these sales in calculating normal value, under Article 2.1, because, according to it, it is implicit in that Article that the exporter must be the seller in order that a sales transaction may properly be used to calculate normal value and this was not the case here. The Appellate Body, reversing the Panel’s finding, considered that Article 2.1 is silent in that respect and that, provided all four explicit conditions (see paragraph 12 above) in Article 2.1 are satisfied, the identity of the “seller of the ‘like product’ is not a ground for precluding the use of a downstream sales transaction when calculating normal value”. However, the Appellate Body stressed that the identity of the seller is not irrelevant when calculating normal value since it may affect comparability, although that aspect is taken care by Article 2.4:

“The text of Article 2.1 is, however, silent as to who the parties to relevant sales transactions should be. Thus, Article 2.1 does not expressly mandate that the sale be made by the exporter for whom a margin of dumping is being calculated. Nor does Article 2.1 expressly preclude that relevant sales transactions might be made downstream, between affiliates of the exporter and independent buyers. In our view, provided that all of the explicit conditions in Article 2.1 of the Anti-Dumping Agreement are satisfied, the identity of the seller of the ‘like product’ is not a ground for precluding the use of a downstream sales transaction when calculating normal value. In short, we see no reason to read into Article 2.1 an additional condition that is not expressed.

We do not mean to suggest that the identity of the seller is irrelevant in calculating normal value under Article 2.1 of the Anti-Dumping Agreement. However, to ensure that prices are ‘comparable’, the Anti-Dumping Agreement provides a mechanism, in Article 2.4, which allows investigating authorities to take full account of the fact, as appropriate, that a relevant sale was not made by the exporter or producer itself, but was made by another party …

…

… the use of downstream sales prices may necessitate the provision of appropriate ‘allowances’, under Article 2.4, which take into account any differences demonstrated to affect price comparability. We will explore this issue further below.”(16)

(b) Sales “in the ordinary course of trade”

(i) Definition of sales “in the ordinary course of trade”

14. In US — Hot-Rolled Steel, the Appellate Body confirmed that the Anti-Dumping Agreement does not define the term “in the ordinary course of trade”.(17) In this dispute, Japan, the complainant, had agreed with the definition of this term given by the United States authorities, namely: “[g]enerally, sales are in the ordinary course of trade if made under conditions and practices that, for a reasonable period of time prior to the date of sale of the subject merchandise, have been normal for sales of the foreign like product.”(18) The Appellate Body considered that for the purpose of the appeal, it was content with that definition.(19)

15. The Appellate Body in US — Hot-Rolled Steel, when looking into the meaning of “sales in the ordinary course of trade” under Article 2.1 of the Anti-Dumping Agreement, noted that Article 2.2.1 does provide for a method to determine whether “sales below cost” are “in the ordinary course of trade”. However, the Appellate Body considered that the said provision does not purport to exhaust the range of methods for determining whether sales are “in the ordinary course of trade” and it does not cover the more specific issue of sales between affiliated parties:

“We note that Article 2.2.1 of the Anti-Dumping Agreement itself provides for a method for determining whether sales below cost are ‘in the ordinary course of trade’. However, that provision does not purport to exhaust the range of methods for determining whether sales are ‘in the ordinary course of trade’, nor even the range of possible methods for determining whether low-priced sales are ‘in the ordinary course of trade’. Article 2.2.1 sets forth a method for determining whether sales between any two parties are ‘in the ordinary course of trade’; it does not address the more specific issue of transactions between affiliated parties. In transactions between such parties, the affiliation itself may signal that sales above cost, but below the usual market price, might not be in the ordinary course of trade. Such transactions may, therefore, be the subject of special scrutiny by the investigating authorities.”(20)

(ii) Investigating authorities’ discretion under Article 2.1

16. The Appellate Body in US — Hot-Rolled Steel noted that the investigating authorities’ discretion under Article 2.1 to determine how to avoid distortions in the normal value should be exercised in a even-handed way that is fair to all parties:

“Although we believe that the Anti-Dumping Agreement affords WTO Members discretion to determine how to ensure that normal value is not distorted through the inclusion of sales that are not ‘in the ordinary course of trade’, that discretion is not without limits. In particular, the discretion must be exercised in an even-handed way that is fair to all parties affected by an anti-dumping investigation. If a Member elects to adopt general rules to prevent distortion of normal value through sales between affiliates, those rules must reflect, even-handedly, the fact that both high and low-priced sales between affiliates might not be ‘in the ordinary course of trade’.”(21)

(iii) Sales not in the ordinary course of trade

Purpose of excluding sales not in the ordinary course of trade

17. In US — Hot-Rolled Steel, the Appellate Body explained that the exclusion of sales not in the ordinary course of trade from the calculation of the normal value is mandated by Article 2.1 in order to ensure that the normal value is indeed “normal”:

“Article 2.1 requires investigating authorities to exclude sales not made ‘in the ordinary course of trade’, from the calculation of normal value, precisely to ensure that normal value is, indeed, the ‘normal’ price of the like product, in the home market of the exporter. Where a sales transaction is concluded on terms and conditions that are incompatible with ‘normal’ commercial practice for sales of the like product, in the market in question, at the relevant time, the transaction is not an appropriate basis for calculating ‘normal’ value.”(22)

Prices above or below the ordinary course of trade price

18. In US — Hot-Rolled Steel, Japan had challenged the so-called “arm’s length” test which allowed the United States authorities to automatically disregard the sales of a given exporter to individual affiliated parties as not being in the ordinary course of trade when the weighted average selling price to that affiliated party is below 99.5 percent of the weighted average price of sales to all non-affiliated parties. Japan claimed that the application of this test was inconsistent with Article 2.1 of the Anti-Dumping Agreement because, first, the test excluded only low-priced affiliated sales, thereby inflating normal value, and, second, the test operated on the basis of an arbitrary threshold that did not take account of usual variation of prices in the marketplace. The Panel found that the application of the 99.5 percent test “does not rest on a permissible interpretation of the term ‘sales in the ordinary course of trade’.”(23) The Appellate Body upheld the Panel’s finding, although it followed a different reasoning.(24)

19. The Appellate Body in US — Hot-Rolled Steel considered that determining “whether a sales price is higher or lower than the ‘ordinary course’ price is not simply a question of comparing prices” and that the other terms and conditions of the transaction must be taken into account:

“We note that determining whether a sales price is higher or lower than the ‘ordinary course’ price is not simply a question of comparing prices. Price is merely one of the terms and conditions of a transaction. To determine whether the price is high or low, the price must be assessed in light of the other terms and conditions of the transaction. Thus, the volume of the sales transaction will affect whether a price is high or low. Or, the seller may undertake additional liability or responsibilities in some transactions, for instance for transport or insurance. These, and a number of other factors, may be expected to affect an assessment of the price.”(25)

20. The Appellate Body in US — Hot-Rolled Steel further considered that nothing excludes that, even in the absence of any common ownership, “a sales transaction might not be ‘in the ordinary course of trade’, either because the sales price is higher than the ‘ordinary course’ price, or because it is lower than that price”:

“Clearly, the lower the degree of common ownership, implying common control, between the parties to a sales transaction, the less likely it is that the transaction will not be ‘in the ordinary course of trade’. However, even where the parties to a sales transaction are entirely independent, a transaction might not be ‘in the ordinary course of trade’.(26) In this appeal, we do not need to define all the circumstances in which transactions might not be ‘in the ordinary course of trade’. It suffices to recognize that, as between affiliates, a sales transaction might not be ‘in the ordinary course of trade’, either because the sales price is higher than the ‘ordinary course’ price, or because it is lower than that price.”(27)

Scope of the investigating authorities’ duties under Article 2.1

21. The Appellate Body in US — Hot-Rolled Steel described the duties of the investigating authorities under Article 2.1 as follows:

“In our view, the duties of investigating authorities, under Article 2.1 of the Anti-Dumping Agreement, are precisely the same, whether the sales price is higher or lower than the ‘ordinary course’ price, and irrespective of the reason why the transaction is not ‘in the ordinary course of trade’. Investigating authorities must exclude, from the calculation of normal value, all sales which are not made ‘in the ordinary course of trade’. To include such sales in the calculation, whether the price is high or low, would distort what is defined as ‘normal value’.

In view of the many different types of transaction not ‘in the ordinary course of trade’ — some including affiliated parties, others not; some including high prices, others low prices; some including prices below cost, others not — investigating authorities need not, under the Anti-Dumping Agreement, scrutinize, according to identical rules, each and every category of sale that is potentially not ‘in the ordinary course of trade’.”(28)

Sales between affiliated companies

22. In US — Hot-Rolled Steel, the Appellate Body upheld the Panel’s findings (albeit for different reasons) that the application by the United States authorities of its 99.5 per cent test to determine whether the sales between affiliated companies were in the ordinary course of trade did not rest upon a permissible interpretation of Article 2.1. See paragraphs 18–20 above.

23. In US — Hot-Rolled Steel, the United States authorities, in calculating the normal value, discarded certain sales by exporters to their affiliates because these sales were not “in the ordinary course of trade”. The authorities had replaced the discarded sales with downstream sales of the product, transacted between the affiliate and the first independent buyer, which had been made “in the ordinary course of trade”. See paragraph 13 above.

(c) Request for information

24. In Guatemala — Cement II, the Panel rejected Mexico’s argument that the request for cost data was not justified under Articles 2.1 and 2.2 because the application did not contain any allegation that Mexican producers were selling below cost, and stated that “[n]othing in those provisions prevents an investigating authority from requesting cost information, even if the applicant does not allege sales below cost.”(29)

(d) Relationship with other paragraphs of Article 2

(i) Article 2.2.1

25. See paragraph 15 above.

(ii) Article 2.4

26. See paragraph 13 above.

3. Article 2.2

(a) Request for cost information

27. With respect to the request for cost information by investigating authorities, see paragraph 24 above.

(b) Article 2.2.1

28. In US — Hot-Rolled Steel, the Appellate Body, when looking into the meaning of “sales in the ordinary course of trade” under Article 2.1, noted that Article 2.2.1 of the Anti-Dumping Agreement “itself provides for a method for determining whether sales below cost are ‘in the ordinary course of trade’. However, that provision does not purport to exhaust the range of methods for determining whether sales are ‘in the ordinary course of trade’, nor even the range of possible methods for determining whether low-priced sales are ‘in the ordinary course of trade’.” See paragraph 15 above.

(i) Article 2.2.1.1

Cost data requirements or elements

29. The Panel on US — DRAMS addressed Korea’s claim that the United States’ authority had acted inconsistently with the first sentence of Article 2.2.1.1 by disregarding cost data which met with the two requirements set forth in the proviso of that Article, namely, “in accordance with generally accepted accounting principles” and “reasonably reflect costs”. The Panel considered that the first sentence is only applicable to “records kept by the exporter or producer under investigation”, and thus refused to apply this Article to cost data prepared by an outside consultant on behalf of the producer.(30)

30. In Egypt — Steel Rebar, the Panel noted that both Articles 2.2.1.1 and 2.2.2 “emphasize two elements, first, that cost of production is to be calculated based on the actual books and records maintained by the company in question so long as these are in keeping with generally accepted accounting principles but that second, the costs to be included are those that reasonably reflect the costs associated with the production and sale of the product under consideration”. (31)

Positive obligations on investigating authorities

31. The Panel on US — Lumber V considered that Article 2.2.1.1 contained only a limited obligation to base the cost on the records of the exporter or producer under investigation under certain circumstances. The Panel was of the view that Article 2.2.1.1 does not require that costs be calculated in accordance with Generally Accepted Accounting Principles (GAAP) nor that they reasonably reflect the costs associated with the production and sale of the product under consideration:

“In our view, Article 2.2.1.1 imposes certain positive obligations on investigating authorities, including the obligation to calculate costs on the basis of records kept by the exporter or producer under investigation and to consider all available evidence on the proper allocation of costs. Neither of these obligations is absolute, however, as in both cases the obligations apply only if (‘provided’) certain conditions are met. The role of these conditions is therefore not to impose positive obligations on Members, but to set forth the circumstances under which certain positive obligations do or do not apply. Thus, Article 2.2.1.1 does not in our view require that costs be calculated in accordance with GAAP nor that they reasonably reflect the costs associated with the production and sale of the product under consideration. Rather, it simply requires that costs be calculated on the basis of the exporter or producer’s records, in so far as those records are in accordance with GAAP and reasonably reflect the costs associated with the production and sale of the product under consideration. Similarly, Article 2.2.1.1 does not require that all allocations made by an investigating authority have been historically utilised by the exporter or producer; rather it simply provides that investigating authorities must consider all available evidence on the proper allocation of costs, including that made available by respondents, insofar as such allocations have been historically utilised by the exporter or producer. Bearing this in mind, we shall examine Canada’s arguments relating to Article 2.2.1.1.”(32)

Consider all available evidence on the proper allocation of costs

32. The Appellate Body on US — Lumber V considered that the requirement to consider all available evidence on the proper allocation of costs may in certain circumstances require the authorities to compare advantages and disadvantages of alternative cost allocation methodologies:

“In our view, the parameters of the obligation to ‘consider all available evidence’ will vary case-by-case. It may well be that, in the light of the facts of a particular case, the requirement to ‘consider all available evidence’ may be satisfied by the investigating authority without comparing allocation methodologies or aspects thereof. However, in other instances — such as where there is compelling evidence available to the investigating authority that more than one allocation methodology potentially may be appropriate to ensure that there is a proper allocation of costs — the investigating authority may be required to ‘reflect on’ and ‘weigh the merits of’ evidence that relates to such alternative allocation methodologies, in order to satisfy the requirement to ‘consider all available evidence’. Thus, although the second sentence of Article 2.2.1.1 does not, as a general rule, require investigating authorities to compare allocation methodologies to assess their respective advantages and disadvantages in each and every case, there may be particular instances in which the investigating authority may be required to compare them in order to satisfy the explicit requirement of the second sentence of Article 2.2.1.1 to ‘consider all available evidence on the proper allocation of costs’.”(33)

Burden of proof

33. Referring to EC — Hormones(34), the Panel on US — DRAMS noted that the burden of establishing a prima facie case of inconsistency with Article 2.2.1.1 was on the complaining party.(35)

(i) Amounts based on actual data pertaining to production and sales of the like product

34. The Panel on US — Lumber V was of the view that amounts for general and administrative expenses pertain to the production and sale of the like product unless it can be demonstrated that the product under investigation did not benefit from a particular General and Administrative costs (G&A) cost item(36):

“We next examine the term ‘pertain to’ within the meaning of the chapeau of Article 2.2.2. ‘Pertain’ is defined as ‘1 a relate or have reference to’.(37) In our view, a meaningful interpretation of the term ‘pertain[ing] to’ must take into account the nature of those costs because, as Canada acknowledges, they ‘are not directly attributable to the product under investigation or [to] any particular product’. Thus, it would appear to us that, unless a particular G&A cost can be tied to a particular product manufactured by a company, G&A costs — because normally they cannot be attributed to any particular product but are costs incurred by the company in the production and sale of goods — pertain or relate to all of those goods. Canada’s argument that G&A costs ‘benefit all products that a company (or division within a company) may produce rather than specific products’ supports our view. If G&A costs benefit the production and sale of all goods that a company may produce, they must certainly relate or pertain to those goods, including in part to the product under investigation.”(38)

(ii) Use of low volume sales Selling, General and Administrative costs (SG&A) and profits data in constructing normal value

35. In its report on EC — Tube or Pipe Fittings, the Appellate Body was asked to examine whether an investigating authority must exclude data from low-volume sales when determining the amounts for SG&A and profits under the chapeau of Article 2.2.2, having disregarded such low-volume sales for normal value determination under Article 2.2. The Appellate Body reasoned as follows:

“Examining the text of the chapeau of Article 2.2.2, we observe that this provision imposes a general obligation (‘shall’) on an investigating authority to use ‘actual data pertaining to production and sales in the ordinary course of trade’ when determining amounts for SG&A and profits. Only ‘[w]hen such amounts cannot be determined on this basis’ may an investigating authority proceed to employ one of the other three methods provided in sub-paragraphs (i)–(iii). In our view, the language of the chapeau indicates that an investigating authority, when determining SG&A and profits under Article 2.2.2, must first attempt to make such a determination using the ‘actual data pertaining to production and sales in the ordinary course of trade’. If actual SG&A and profit data for sales in the ordinary course of trade do exist for the exporter and the like product under investigation, an investigating authority is obliged to use that data for purposes of constructing normal value; it may not calculate constructed normal value using SG&A and profit data by reference to different data or by using an alternative method.

As the Panel correctly observed, it is meaningful for the interpretation of Article 2.2.2 that Article 2.2 specifically identifies low-volume sales in addition to sales outside the ordinary course of trade. In contrast to Article 2.2, the chapeau of Article 2.2.2 explicitly excludes only sales outside the ordinary course of trade. The absence of any qualifying language related to low volumes in Article 2.2.2 implies that an exception for low-volume sales should not be read into Article 2.2.2.”(39)

36. The Appellate Body in EC — Tube or Pipe Fittings conluded that it is “significant that Article 2.2.2 specifies the data to be used by an investigating authority when constructing normal value. The text of that provision excludes actual data outside the ordinary course of trade, but does not exclude data from low-volume sales. The negotiators’ express reference to sales outside the ordinary course of trade and to low-volume sales in Article 2.2, and the omission of a reference to low-volume sales in the chapeau of Article 2.2.2, confirms our view that low-volume sales are not excluded from the chapeau of Article 2.2.2 for the calculation of SG&A profits.”(40) Thus, the Appellate Body found that in cases where low-volume sales are in the ordinary course of trade, an investigating authority does not act inconsistently with the chapeau of Article 2.2.2 by including actual data from those sales to derive SG&A and profits for the construction of normal value.

(iii) Priority of options

37. In response to the argument that the order of methodological options for calculating reasonable amount for profit set out in Article 2.2.2 reflects a preference for one option over another, the Panel on EC — Bed Linen, in a finding subsequently not addressed by the Appellate Body, concluded that “the order in which the three options are set out in Article 2.2.2(i)–(iii) is without any hierarchical significance and that Members have complete discretion as to which of the three methodologies they use in their investigations.” (41) The Panel set out the following reasoning:

“Looking first at the text of Article 2.2.2, we see nothing that would indicate that there is a hierarchy among the methodological options listed in subparagraphs (i) to (iii). Of course, they are listed in a sequence, but this is an inherent characteristic of any list, and does not in and of itself entail any preference of one option over others. Moreover, we note that where the drafters intended an order of preference, the text clearly specifies it. … Had the drafters wished to indicate a hierarchy among the three options, surely they would have done so in a manner that made that hierarchy explicit. Certainly, we would have expected something more than simply a numbered listing. Thus, in context, it seems clear to us that the mere order in which the options appear in Article 2.2.2 has no preferential significance.

… Paragraphs (i)–(iii) provide three alternative methods for calculating the profit amount, which, in our view, are intended to constitute close approximations of the general rule set out in the chapeau of Article 2.2.2. These approximations differ from the chapeau rule in that they relax, respectively, the reference to the like product, the reference to the exporter concerned, or both references, spelled out in that rule …

In our view, there is no basis on which to judge which of these three options is ‘better’. Certainly, there were differing views during the negotiations as to how this issue was to be resolved, and there is no specific language in the Agreement to suggest that the drafters considered one option preferable to the others. Given, as explained above, that each of the three options is in some sense ‘imperfect’ in comparison with the chapeau methodology, there is, in our opinion, no meaningful way to judge which option is less imperfect — or of greater authority — than another and, thus, no obvious basis for a hierarchy. And it is, in our view, for the drafters of an Agreement to set out a hierarchy or order of preference among admittedly imperfect approximations of a preferred result, and not for a panel to impose such a choice where it is not apparent from the text.”(42)

(iv) Relationship with Article 2.2.2

38. See paragraph 30 above.

(v) Article 2.2.2(i) — “same general category of products”

39. In Thailand — H-Beams, in a finding not reviewed by the Appellate Body, the Panel rejected Poland’s argument that the Thai authority had, for the purpose of calculating profit in constructed normal value, adopted too narrow a definition of the term “same general category of products”. The Panel stated:

“[W]e note that the text of Article 2.2.2 (i) simply refers without elaboration to ‘the same general category of products’ produced by the producer or exporter under investigation. Thus, the text of this subparagraph provides no precise guidance as to the required breadth or narrowness of the product category, and therefore provides no support for Poland’s argument that a broader rather than a narrower definition is required.”(43)

40. The Panel on Thailand — H-Beams went on to explain the contextual bases for its interpretation of Article 2.2.2(i) quoted in paragraph 39 above. The Panel first opined that the context of Article 2.2.2 (i) supports a narrow rather than a broad interpretation of the term “same general category of products”:

“We do find a certain amount of guidance in other provisions of Article 2.2.2, in particular its chapeau and its overall structure, however. In particular, we note that, in general, Article 2.2 and Article 2.2.2 concern the establishment of an appropriate proxy for the price ‘of the like product in the ordinary course of trade in the domestic market of the exporting country’ when that price cannot be used. As such, as the drafting of the provisions makes clear, the preferred methodology which is set forth in the chapeau is to use actual data of the exporter or producer under investigation for the like product. Where this is not possible, subparagraphs (i) and (ii) respectively provide for the database to be broadened, either as to the product (i.e., the same general category of products produced by the producer or exporter in question) or as to the producer (i.e., other producers or exporters subject to investigation in respect of the like product), but not both. Again this confirms that the intention of these provisions is to obtain results that approximate as closely as possible the price of the like product in the ordinary course of trade in the domestic market of the exporting country.

This context indicates to us that the use under subparagraph (i) of a narrower rather than a broader ‘same general category of products’ certainly is permitted. Indeed, the narrower the category, the fewer products other than the like product will be included in the category, and this would seem to be fully consistent with the goal of obtaining results that approximate as closely as possible the price of the like product in the ordinary course of trade in the domestic market of the exporting country.”(44)

41. The Panel on Thailand — H-Beams found additional contextual support in Article 3.6 for its finding that the term “same general category of products” under Article 2.2.2(i) permits a narrower rather than a broader approach:

“Additional contextual support can be found in Article 3.6 (a provision related to data concerning injury), which provides that when available data on ‘criteria such as the production process, producers’ sales and profits’ do not permit the separate identification of production of the like product, ‘the effects of the dumped imports shall be assessed by the examination of the production of the narrowest group or range of products, which includes the like product, for which the necessary information can be provided’ (emphasis supplied). Although this provision concerns information relevant to injury rather than dumping, and although we do not mean to suggest that use of the narrowest possible category including the like product is required under Article 2.2.2(i), in our view Article 3.6 provides contextual support for the conclusion that use of a narrow rather than a broader category is permitted.

We note Poland’s argument that a broader category is more likely than a narrower one to yield ‘representative’ results (by which we presume Poland to mean representative of the price of the like product in the ordinary course of trade in the domestic market of the exporting country), but we believe that as a matter of logic the opposite more often is likely to be true. The broader the category, the more products other than the like product will be included, and thus in our view the more potential there will be for the constructed normal value to be unrepresentative of the price of the like product. We therefore disagree with Poland that Article 2.2.2(i) requires the use of broader rather than narrower categories, and believe to the contrary that the use even of the narrowest general category that includes the like product is permitted.”(45)

(vi) Article 2.2.2(ii) — “weighted average” and data from “other exporters or producers”

42. In EC — Bed Linen, the Appellate Body reversed the Panel’s finding under Article 2.2.2(ii) that the existence of data for more than one other exporter or producer is not a necessary prerequisite for application of the approach using “weighted average” in calculating the amount for administrative, selling and general costs (“SG&A”) to determine the constructed normal value of subject products. The Appellate Body stated:

“In our view, the phrase ‘weighted average’ in Article 2.2.2(ii) precludes, in this particular provision, understanding the phrase ‘other exporters or producers’ in the plural as including the singular case. To us, the use of the phrase ‘weighted average’ in Article 2.2.2(ii) makes it impossible to read ‘other exporters or producers’ as ‘one exporter or producer’. First of all, and obviously, an ‘average’ of amounts for SG&A and profits cannot be calculated on the basis of data on SG&A and profits relating to only one exporter or producer. Moreover, the textual directive to ‘weight’ the average further supports this view because the ‘average’ which results from combining the data from different exporters or producers must reflect the relative importance of these different exporters or producers in the overall mean. In short, it is simply not possible to calculate the ‘weighted average’ relating to only one exporter or producer. Indeed, we note that, at the oral hearing in this appeal, the European Communities conceded that the phrase ‘weighted average’ envisages a situation where there is more than one exporter or producer.

The requirement to calculate a ‘weighted average’ in Article 2.2.2(ii) is, in our view, the key to interpreting that provision. It is indispensable to the calculation method set forth in this provision, and, thus, it is indispensable to the entire provision — which deals only with the mechanics of that calculation. We disagree with the Panel that ‘the concept of weighted averaging is relevant only when there is information from more than one other producer or exporter available to be considered.’ (emphasis in the original) We see no justification, textual or otherwise, for concluding that amounts for SG&A and profits are to be determined on the basis of the weighted average some of the time but not all of the time. In so interpreting Article 2.2.2(ii), the Panel, in effect, reads the requirement of calculating a ‘weighted average’ out of the text in some circumstances. In those circumstances, this would substantially empty the phrase ‘weighted average’ of meaning. (46)

In our view, then, the use of the phrase ‘weighted average’, combined with the use of the words ‘amounts’ and ‘exporters or producers’ in the plural in the text of Article 2.2.2(ii), clearly anticipates the use of data from more than one exporter or producer. We conclude that the method for calculating amounts for SG&A and profits set out in this provision can only be used if data relating to more than one other exporter or producer is available.”(47)

(vii) Article 2.2.2(ii) — production and sales amounts “incurred and realized”

43. In EC — Bed Linen, the Appellate Body reversed the Panel’s conclusion that “an interpretation of Article 2.2.2(ii) under which sales not in the ordinary course of trade are excluded from the determination of the profit amount to be used in the calculation of a constructed normal value is permissible”.(48) The Appellate Body emphasized that Article 2.2.2(ii) refers to “actual amounts incurred and realized by other exporters and producers” and concluded that, in the light of this wording, in the calculation of weighted average all of the actual amounts have to be included, regardless of whether the underlying sales were made in the ordinary course of trade or not:

“Here, we note especially that Article 2.2.2(ii) refers to ‘the weighted average of the actual amounts incurred and realized by other exporters or producers’. (emphasis added) In referring to ‘the actual amounts incurred and realized’, this provision does not make any exceptions or qualifications. In our view, the ordinary meaning of the phrase ‘actual amounts incurred and realized’ includes the SG&A actually incurred, and the profits or losses actually realized(49) by other exporters or producers in respect of production and sales of the like product in the domestic market of the country of origin. There is no basis in Article 2.2.2(ii) for excluding some amounts that were actually incurred or realized from the ‘actual amounts incurred or realized’. It follows that, in the calculation of the ‘weighted average’, all of ‘the actual amounts incurred and realized’ by other exporters or producers must be included, regardless of whether those amounts are incurred and realized on production and sales made in the ordinary course of trade or not. Thus, in our view, a Member is not allowed to exclude those sales that are not made in the ordinary course of trade from the calculation of the ‘weighted average’ under Article 2.2.2(ii).”(50)

44. The Appellate Body in EC — Bed Linen also discussed the first sentence of the chapeau of Article 2.2.2 as part of the context supporting its interpretation of Article 2.2.2(ii) quoted in paragraph 43 above. The Appellate Body stated:

“In contrast to Article 2.2.2(ii), the first sentence of the chapeau of Article 2.2.2 refers to ‘actual data pertaining to production and sales in the ordinary course of trade’. (emphasis added) Thus, the drafters of the Anti-Dumping Agreement have made clear that sales not in the ordinary course of trade are to be excluded when calculating amounts for SG&A and profits using the method set out in the chapeau of Article 2.2.2.

The exclusion in the chapeau leads us to believe that, where there is no such explicit exclusion elsewhere in the same Article of the Anti-Dumping Agreement, no exclusion should be implied. And there is no such explicit exclusion in Article 2.2.2(ii). Article 2.2.2(ii) provides for an alternative calculation method that can be employed precisely when the method contemplated by the chapeau cannot be used. Article 2.2.2(ii) contains its own specific requirements. On their face, these requirements do not call for the exclusion of sales not made in the ordinary course of trade. Reading into the text of Article 2.2.2(ii) a requirement provided for in the chapeau of Article 2.2.2 is not justified either by the text or by the context of Article 2.2.2(ii).”(51)

(viii) Article 2.2.2(ii) — should “weighted” average be based on the value or the volume of sales?

45. The Panel on EC — Bed Linen (Article 21.5 — India) rejected India’s claim that the weighting of averages under Article 2.2.2 (ii) was to be perfomed on the basis of sales volume rather than value data. According to the Panel,

“It is clear from the text of Article 2.2.2(ii) that the amounts for SG&A and for profits to be used in constructing normal value must be weighted averages. However, nothing in the text specifies the factor to be used in calculating those weighted averages. There is clearly no specific direction requiring that the averages be weighted on the basis of volume, rather than value. Article 2.2.2(ii) is simply silent on this issue. Article 2.2.2 (ii) does not specify the factor, volume or value, to be used in calculating weighted averages.” (52)

46. The Panel on EC — Bed Linen (Article 21.5 — India) further explained that, in its view, “either volume or value may be relevant in the context of Article 2.2.2(ii), and both are ‘neutral’ in the sense that the weighted average will reflect the relative importance of the companies with respect to that factor”.(53) According to the Panel, “the fact that the choice of the factor used in calculating the weighted average will affect the outcome is simply irrelevant to the question whether Article 2.2.2(ii) requires the use of one volume rather than value as the weighting factor.”(54)

(ix) No separate “reasonability” test

47. The Panel on EC — Bed Linen, in a finding subsequently not addressed by the Appellate Body, rejected the argument by India that “the results of a proper calculation under Article 2.2.2(ii) are subject to a separate test of ‘reasonability’ before they may be used in constructing a normal value for other producers” (55). The Panel was unable to find a basis for such a separate reasonability test in the wording of Article 2.2.2:

“The text … indicates that the methodologies set out in Article 2.2.2 are outlined ‘for the purpose’ of calculating a reasonable profit amount pursuant to Article 2.2. There is no specific language establishing a separate reasonability test, or indicating how such a test should be conducted. In these circumstances, we consider that there is no textual basis for such a requirement. Thus, the ordinary meaning of the text indicates that if one of the methods of Article 2.2.2 is properly applied, the results are by definition ‘reasonable’ as required by Article 2.2.

Further, we note that Article 2.2.2(iii) provides for the use of ‘any other reasonable method’, without specifying such method, subject to a cap, defined as ‘the profit normally realized by other exporters or producers on sales of products of the same general category in the domestic market of the country of origin’. To us, the inclusion of a cap where the methodology is not defined indicates that where the methodology is defined, in subparagraphs (i) and (ii), the application of those methodologies yields reasonable results. If those methodologies did not yield reasonable results, presumably the drafters would have included some explicit constraint on the results, as they did for subparagraph (iii).

Thus, we conclude that the text indicates that, if a Member bases its calculations on either the chapeau or paragraphs (i) or (ii), there is no need to separately consider the reasonability of the profit rate against some benchmark. In particular, there is no need to consider the limitation set out in paragraph (iii). That limitation is triggered only when a Member does not apply one of the methods set out in the chapeau or paragraphs (i) and (ii) of Article 2.2.2. Indeed, it is arguably precisely because no specific method is outlined in paragraph (iii) that the limitation on the profit rate exists in that provision.”(56)

48. Similarly to the Panel on EC — Bed Linen, the Panel on Thailand — H-Beams also considered that no separate “reasonability” test is required under Article 2.2.2, and rejected Poland’s argument that the results of applying any of the specified methodologies are at best rebuttably presumed to be reasonable. The Panel stated:

“We find no trace in the texts of the relevant provisions of such a rebuttable presumption, however. To the contrary, the ordinary meaning of the text seems rather to indicate that, if one of the methodologies is applied, the result is by definition reasonable. First, as noted, the phrase ‘for the purpose of paragraph 2’ is without qualification in the text. In our view, this phrase is straightforward and means that Article 2.2.2 gives the specific instructions as to how to fulfil the basic but unelaborated requirement in Article 2.2 to use no more than a ‘reasonable’ amount for profit.

Second, we note that the chapeau of Article 2.2.2 provides that where the methodology in the chapeau ‘cannot’ be used, one of the methodologies in subparagraphs (i), (ii) or (iii) ‘may’ be used. Poland argues that the word ‘may’ only provides for the possibility of using such methodologies and implies that any results derived thereby would be subject to a reasonability test arising under Article 2.2. We disagree, as in our view the word ‘may’ constitutes authorization to use the methodologies in the subparagraphs where the methodology in the chapeau, which is the preferred methodology, ‘cannot’ be used. We note that the text of Article 2.2.2 establishes no hierarchy among the subparagraphs and that there is no disagreement between the parties concerning this issue.”(57)

49. The Panel on Thailand — H-Beams, similarly to the Panel on EC — Bed Linen, went on to find that the existence of a “cap” under subparagraph (iii) of Article 2.2.2. with respect to “any other reasonable method” implied that the methodologies under subparagraphs (i) and (ii) ipso facto yielded “reasonable” results, such that no separate constraint existed in respect of these paragraphs.(58) The Panel, in a finding subsequently not reviewed by the Appellate Body, then also noted the requirement to use “actual data” under the Article 2.2.2 chapeau and subparagraphs (i) and (ii):

“We note also the requirement in the chapeau of Article 2.2.2 as well as in subparagraphs (i) and (ii) that actual data be used. In our view, the notion of a separate reasonability test is both illogical and superfluous where the Agreement requires the use of specific types of actual data. That is, where actual data are used and the other requirements of the relevant provision(s) are fulfilled (e.g., that the ‘same general category of products’ is defined in a permissible way where 2.2.2(i) is applied), a correct or accurate result is obtained, and the requirement to use actual data is itself the mechanism that ensures reasonability in the sense of Article 2.2 of that (correct) result. By contrast, under subparagraph (iii) where no specific methodology or data source is required, and the use of ‘any other reasonable method’ is permitted, the provision itself contains what is in effect a separate reasonability test, namely the cap on the profit amount based on the actual experience of other exporters or producers. Thus, in our view, Article 2.2.2’s requirement that actual data be used (and its establishment of a cap where this is not the case) are intended precisely to avoid the outcome that Poland seeks, namely subjective judgements by national authorities as to the ‘reasonability’ of given amounts used in constructed value calculations.”(59)

(d) Relationship with other paragraphs of Article 2

50. In Egypt — Steel Rebar, the Panel indicated that, in its view, what might be necessary to take into account by way of due allowance in a particular investigation in order to comply with the obligation to ensure a fair comparison under Article 2.4 could not be limited by the simplistic characterisation of a normal value as being one arrived at by way of a construction under Article 2.2:

“[W]e do not think that the construction of a normal value under Article 2.2 precludes consideration of the making of various adjustments as between that normal value and the export price with which it is to be compared. A constructed normal value is, in effect, a notional price, ‘built up’ by adding costs of production, administrative, selling and other costs, and a profit. In any given case, such a built-up price might or might not reflect credit costs. Thus, what might be necessary to take into account by way of due allowance in a particular investigation in order to comply with the obligation to ensure a fair comparison under Article 2.4 cannot be limited by the simplistic characterisation of a normal value as being one arrived at by way of a construction under Article 2.2.”(60)

51. The Panel on EC — Tube or Pipe Fittings found that the definition of “like product” in Article 2.6 governs how an investigating authority identifies the scope of the “like product” for the purposes of the investigation and of the Agreement. The Panel considered that, since the chapeau of Article 2.2.2 requires the use of actual data from all relevant sales of the like product, “actual data from relevant transactions relating to sales of the ‘like product’ — as a whole — may be taken into account to construct normal value. There is no provision to the effect that constructed normal value is to be based only on a limited subset of data relating to sales of certain selective product types falling within the definition of like product, but excluding data relating to sales of other such types.”(61)

4. Article 2.3

52. In US — Stainless Steel, the Panel explained the status of paragraph 3 in Article 2. See paragraph 11 above.

5. Article 2.4

(a) First sentence

(i) Fair comparison of export price and normal value

53. In Egypt — Steel Rebar, the Panel considered that “Article 2.4 in its entirety, including its burden of proof requirement, has to do with ensuring a fair comparison, through various adjustments as appropriate, of export price and normal value”.(62) The Panel indicated that the ordinary meaning of this provision confirms that it has to do with the nature of the comparison of export price and normal value. In the Panel’s view, “the immediate context of this provision, namely Articles 2.4.1 and 2.4.2 confirms that Article 2.4 and in particular its burden of proof requirement, applies to … the calculation of the dumping margin”. The Panel thus found that this provision did not apply to the investigating authority’s establishment of normal value as such:

“Article 2.4, on its face, refers to the comparison of export price and normal value, i.e., the calculation of the dumping margin, and in particular, requires that such a comparison shall be ‘fair’. A straightforward consideration of the ordinary meaning of this provision confirms that it has to do not with the basis for and basic establishment of the export price and normal value (which are addressed in detail in other provisions)(63), but with the nature of the comparison of export price and normal value. First, the emphasis in the first sentence is on the fairness of the comparison. The next sentence, which starts with the words ‘[t]his comparison’, clearly refers back to the ‘fair comparison’ that is the subject of the first sentence. The second sentence elaborates on considerations pertaining to the ‘comparison’, namely level of trade and timing of sales on both the normal value and export price sides of the dumping margin equation. The third sentence has to do with allowances for ‘differences which affect price comparability’, and provides an illustrative list of possible such differences. The next two sentences have to do with ensuring ‘price comparability’ in the particular case where a constructed export price has been used. The final sentence, where the reference to burden of proof at issue appears, also has to do with ‘ensur[ing] a fair comparison’. In particular, the sentence provides that when collecting from the parties the particular information necessary to ensure a fair comparison, the authorities shall not impose an unreasonable burden of proof on the parties.

The immediate context of this provision, namely Articles 2.4.1 and 2.4.2 confirms that Article 2.4 and in particular its burden of proof requirement, applies to the comparison of export price and normal value, that is, the calculation of the dumping margin. Article 2.4.1 contains the relevant provisions for the situation where ‘the comparison under paragraph 4 requires a conversion of currencies’ (emphasis added). Article 2.4.2 specifically refers to Article 2.4 as ‘the provisions governing fair comparison’, and then goes on to establish certain rules for the method by which that comparison is made (i.e., the calculation of dumping margins on a weighted-average to weighted-average or other basis).”(64)

(ii) Relationship with other sentences

54. In US — Stainless Steel, having found a violation of the third and fourth sentence of Article 2.4 in respect of certain allowances, the Panel considered that it was “not … necessary to examine Korea’s claims that the United States’ treatment of bad debt breached a more general ‘fair comparison’ requirement under Article 2.4 of the AD Agreement.”(65)

(b) Second sentence

(i) “sales made at as nearly as possible the same time”

55. The Panel on US — Stainless Steel rejected the United States argument that the “same time” requirement of Article 2.4 implies a preference for shorter rather than longer averaging periods when calculating the dumping margin pursuant to the weighted average/weighted average method in Article 2.4.2, first sentence. See paragraph 86 below.

(i) “in each case, on its merits”

56. In Argentina — Ceramic Tiles, the Panel analysed the meaning of the requirement to make “due allowance in each case, on its merits” for differences in physical characteristics affecting price comparability. The Panel concluded that this requirement “means at a minimum that the authority has to evaluate identified differences in physical characteristics” and not only the most important ones:

“Article 2.4 places the obligation on the investigating authority to make due allowance, in each case on its merits, for differences which affect price comparability, including differences in physical characteristics. The last sentence of Article 2.4 provides that the authorities shall indicate to the parties in question what information is necessary to ensure a fair comparison. We believe that the requirement to make due allowance for such differences, in each case on its merits, means at a minimum that the authority has to evaluate identified differences in physical characteristics to see whether an adjustment is required to maintain price comparability and to ensure a fair comparison between normal value and export price under Article 2.4 of the AD Agreement, and to adjust where necessary.

…

… We do not agree with Argentina’s view that Article 2.4, through the qualifying language that due allowance shall be made ‘in each case’ ‘on its merits’, permits an investigating authority to adjust only for the most important of the physical differences that affect price comparability, even if making the remaining adjustments would have been, as the parties agree, complex. The DCD chose not to conduct a model-by-model comparison and it was then left to find other means to account for the remaining physical differences affecting price comparability. It did not do so.”(66)

57. In Egypt — Steel Rebar, the Panel read Article 2.4 as explicitly requiring a fact-based, case-by-case analysis of differences that affect price comparability:

“[W]e read Article 2.4 as explicitly requiring a fact-based, case-by-case analysis of differences that affect price comparability. In this regard, we take note in particular of the requirement in Article 2.4 that ‘[d]ue allowance shall be made in each case, on its merits, for differences which affect price comparability’ (emphasis added). We note as well that in addition to an illustrative list of possible such differences, Article 2.4 also requires allowances for ‘any other differences which are also demonstrated to affect price comparability’ (emphasis added). Finally, we note the affirmative information-gathering burden on the investigating authority in this context, that it ‘shall indicate to the parties in question what information is necessary to ensure a fair comparison and shall not impose an unreasonable burden of proof on those parties’ (emphasis added). In short, where it is demonstrated by one or another party in a particular case, or by the data itself that a given difference affects price comparability, an adjustment must be made. In identifying to the parties the data that it considers would be necessary to make such a demonstration, the investigating authority is not to impose an unreasonable burden of proof on the parties. Thus, the process of determining what kind or types of adjustments need to be made to one or both sides of the dumping margin equation to ensure a fair comparison, is something of a dialogue between interested parties and the investigating authority, and must be done on a case-by-case basis, grounded in factual evidence.”(67)

58. The Panel on EC — Tube or Pipe Fittings considered that Article 2.4 did not set forth any particular methodology for calculating adjustments and that a Panel could therefore only examine whether the investigating authority acted in an unbiased and even-handed manner when calculating the adjustments made:

“An investigating authority must act in an unbiased, evenhanded manner and must not exercise its discretion in an arbitrary manner. This obligation also applies where an investigating authority confronts practical difficulties and time constraints. We do not find, in Article 2.4, or in any other relevant provision in the Agreement, any specific rules governing the methodology to be applied by an investigating authority in calculating adjustments. In the absence of any precise textual guidance in the Agreement concerning how adjustments are to be calculated, and in the absence of any textual prohibition on the use of any particular methodology adopted by an investigating authority with a view to ensuring a fair comparison, we consider that an unbiased and objective authority could have applied this methodology applied by the European Communities and calculated this adjustment on the basis of the actual data in the record of this investigation. Moreover, Tupy had an opportunity to substantiate its claimed adjustment.”(68)

(ii) “differences which affect price comparability”

59. In US — Hot-Rolled Steel, the Appellate Body ruled that the investigating authorities cannot exclude any differences affecting price comparability from being the object of an allowance:

“Article 2.4 of the Anti-Dumping Agreement provides that, where there are ‘differences’ between export price and normal value, which affect the ‘comparability’ of these prices, ‘[d]ue allowance shall be made’ for those differences. The text of that provision gives certain examples of factors which may affect the comparability of prices: ‘differences in conditions and terms of sale, taxation, levels of trade, quantities, physical characteristics, and any other differences’. However, Article 2.4 expressly requires that ‘allowances’ be made for ‘any other differences which are also demonstrated to affect price comparability.’ (emphasis added) There are, therefore, no differences ‘affect[ing] price comparability’ which are precluded, as such, from being the object of an ‘allowance’.”(69)

60. The Panel on US — Lumber V considered that there is no requirement to adjust for any and all differences but rather only those differences demonstrated to have affected the price comparability:

“We consider that Article 2.4 does not require that an adjustment be made automatically in all cases where a difference is found to exist, but only where — based on the merits of the case — that difference is demonstrated to affect price comparability. An interpretation that an adjustment would have to be made automatically where a difference in physical characteristics is found to exist would render the term ‘which affect price comparability’ nugatory. Further, such an interpretation would make little sense in practice, as not all differences in physical characteristics necessarily affect price comparability “(70)

61. Reflecting further on the meaning of the term comparability in Article 2.4, the Panel on US — Lumber V concluded that an investigating authority must, based on the facts before it, on a case-by-case basis decide whether a certain factor is demonstrated to affect price comparability:

“The identified differences concerning the products sold in the two markets must affect the comparability of normal value and export price for the obligation to make due allowance to apply. Article 2.4 does not define what comparability means, but includes a non-exhaustive list of factors which may affect price comparability. Comparability is a term which, in our view, cannot be defined in the abstract. Rather, an investigating authority must, based on the facts before it, on a case-by-case basis decide whether a certain factor is demonstrated to affect price comparability. We can imagine of situations where although differences exist, they do not affect price comparability. For instance, this could occur where in the exporting country all cars sold are painted in red, while cars exported are all black. The difference is obvious; in fact, it is one of those differences listed in Article 2.4 itself — a difference in physical characteristics. However, there might be no variable cost difference among the two cars because the cost of the paint — whether red or black — might be the same. If instead of differences in cost, we were looking at market value differences, we might reach the same conclusion if, either the seller or the purchaser, would be willing to sell or purchase at the same price, regardless whether the car is red or black.”(71)

(iii) Differences in “terms and conditions of sale”

62. In US — Stainless Steel, the Panel examined Korea’s argument that in violation of the third sentence of Article 2.4, which permits an adjustment “for differences affecting price comparability, including differences in conditions and terms of sales …”, the United States treated export sales which had not been paid because the customer had gone bankrupt later, as “direct selling expenses”, and allocated these direct selling expenses over all United States’ sales. The Panel rejected the United States’ argument that bad debts are expenses directly related to the payment terms of the contract, and stated:

“We do not consider that the phrase ‘differences in conditions and terms of sale’, interpreted in accordance with customary rules of interpretation of public international law, can be understood to encompass differences arising from the unforeseen bankruptcy of a customer and consequent failure to pay for certain sales. In this respect, we note that Article 2.4 refers to the ‘terms and conditions of sale’. Although of course both words — ‘term’ and ‘condition’ — have many meanings, both are commonly used in relation to contracts and agreements. Thus, ‘term’ is defined, inter alia, to mean ‘conditions with regard to payment for goods or services’, while ‘condition’ is defined, inter alia, as ‘a provision in a will, contract, etc., on which the force or effect of the document depends’. Thus, we consider that, read as a whole, the phrase ‘conditions and terms of sale’ refers to the bundle of rights and obligations created by the sales agreement, and ‘differences in conditions and terms of sale’ refers to differences in that bundle of contractual rights and obligations. Thus, to the extent that there are, for example, differences in payment terms in the two markets, a difference in the conditions and terms of sale exists. The failure of a customer to pay is not a condition or term of sale in this sense, however. Rather, non-payment involves a situation where the purchaser has violated the ‘conditions and terms of sale’ by breaching its obligation to pay for the merchandise in question.”(72)

63. The Panel on US — Stainless Steel specifically responded to the United States’ argument that unpaid export sales were to be treated as “direct selling expenses” in distinguishing between “differences in conditions and terms of sale” and the “mode or state of being” of such sales:

“We perceive no textual basis for the United States’ effort to characterize all differences in costs associated with the terms of the contract and expenses directly related to the sale as ‘differences in terms and conditions of sale’. The United States contends that ‘conditions’ of sale can be read in this context to mean the ‘mode or state of being’ of sales, such that ‘differences in conditions and terms of sale’ include the ‘mode or circumstances’ under which sales are made. Assuming this interpretation to be a permissible one, it might allow for adjustments for ‘differences in conditions and terms of sale’ in cases where the contractual provisions governing sales in the two markets were identical but the seller was aware from circumstances existing at the time of sale that those provisions would likely entail different costs.(73) Thus to take an example often cited by the United States in this dispute, a seller might extend identical warranties in different markets or to different customers, knowing in advance that the costs related to those warranties in one market would likely be higher than in the other. Similarly, a seller might extend sales on the same credit terms in two different markets or to two different customers in the awareness that the risk of default — and thus the likely costs associated with the extension of credit — would be higher in one case than in the other. However, we fail to see how the fact that a customer who has purchased on credit subsequently went bankrupt and failed to pay for his purchases could be deemed a ‘circumstance under which sales are made’, at least in a case such as this one where the seller had no knowledge of the precarious financial situation of the purchaser.

We consider that an examination of the context in which the phrase ‘differences in conditions and terms of sale’ is used supports our understanding of the ordinary meaning of this phrase. We recall that Article 2.4 identifies ‘differences in conditions and terms of sale’ as one of several ‘differences which affect price comparability’. Thus, the notion of price comparability informs our interpretation of ‘differences in conditions and terms of sale’. In our view, the requirement to make due allowance for differences that affect price comparability is intended to neutralise differences in a transaction that an exporter could be expected to have reflected in his pricing. A difference that could not reasonably have been anticipated and thus taken into account by the exporter when determining the price to be charged for the product in different markets or to different customers is not a difference that affects the comparability of prices within the meaning of Article 2.4. This reinforces our view that the phrase ‘differences in conditions and terms of sale’ cannot permissibly be interpreted to encompass an unanticipated failure of a customer to pay for certain sales.”(74)

64. Further, the Panel on US — Stainless Steel rejected the United States’ argument that its methodology for the treatment of bad debt was simply a practical way to address differing levels of risks between markets in cases where sales are made on credit. The Panel opined that differences in risk of non-payment might be a difference relevant for the purposes of Article 2.4 and that actual bad debt could be evidence for establishing such different levels of risk of non-payment. However, it found that the United States’ methodology did not base its determination on these factors:

“[W]e agree with the United States that a difference in risk of non-payment between markets that was known at the time of sale might represent a difference for which due allowance could properly be made under Article 2.4. Nor do we preclude that actual bad debt experience during the period of investigation might be evidence relevant to establishing the existence of such a difference.(75) The United States did not however treat actual experience with respect to levels of unpaid sales as evidence of different levels of risk in the two markets in these investigations. Rather, it stated that it was the DOC’s practice to treat bad debt as a direct selling expense when the expense was incurred in respect of the subject merchandise. Thus, even assuming that the US methodology was somehow intended to address differences in risk of non-payment, we do not accept the proposition that the existence of a higher level of non-payment in one market than in another during the period of investigation may be deemed to demonstrate the existence of such differences in risk and thus represent a permissible adjustment for ‘differences in conditions and terms of sale’.(76)”(77)

(d) Fourth sentence

(i) Legal effect

65. In US — Stainless Steel, the United States argued that the fourth sentence of Article 2.4 is not mandatory since it provides that allowances for costs and profits “should” be made in constructing an export price. The Panel agreed that the Anti-Dumping Agreement permits, but does not require such allowances, but opined that a Member may not make allowances other than those authorized by Article 2.4:

“The term ‘should’ in its ordinary meaning generally is non-mandatory, i.e., its use in this sentence indicates that a Member is not required to make allowance for costs and profits when constructing an export price.(78) We believe that, because the failure to make allowance for costs and profits could only result in a higher export price — and thus a lower dumping margin — the AD Agreement merely permits, but does not require, that such allowances be made.(79)

… In our view, that the AD Agreement does not require such allowances does not mean that a Member is free to make any allowances it desires, including allowances not specified in this provision. To the contrary, we view this sentence as providing an authorization to make certain specific allowances. We therefore consider that allowances not within the scope of that authorization cannot be made.(80) If a Member were free to make any additional allowances it desired, there would be no effective disciplines on the methodology for construction of an export price and the provision in question would in our view be reduced to inutility.(81) Thus, we conclude that it would be inconsistent with Article 2.4 of the AD Agreement to make allowances in the construction of the export price that are not within the scope of the authorization found in that Article.

Our conclusion that Article 2.4 contains binding obligations regarding the scope of the permissible allowances that can be made in constructing an export price does not mean that we equate allowances for differences which affect price comparability with allowances relating to the construction of the export price. Rather, the third sentence of Article 2.4 requires due allowance to be made for differences affecting price comparability, while the fourth sentence provides that in the cases referred to in paragraph 3 — i.e., when constructing an export price — allowance for certain costs and profits should also be made. Finally, the fifth sentence of Article 2.4 makes clear that allowances relating to the construction of the export price could in fact reduce price comparability, such that one of several compensating steps should be taken. For all these reasons, it is clear to us that allowances in respect of construction of the export price are separate and distinct from allowances for differences which affect price comparability and are governed by different substantive rules.”(82)

(ii) “costs … incurred between importation and resale”

66. In US — Stainless Steel, the Panel agreed with Korea’s argument that it was inconsistent with Article 2.4 to treat export sales unpaid as a result of the bankruptcy of a customer as direct selling costs, because the related costs were not “incurred between importation and resale” mentioned in the fourth sentence of Article 2.4. The Panel established the “foreseeability” of costs as the decisive factor:

“[W]e note that Article 2.4 uses the word ‘between’. That term is defined to mean, inter alia, ‘[i]n the interval separating two points of time, events, etc.’. Thus, the phrase costs ‘incurred between importation and resale’ in its ordinary meaning is most naturally read to refer to costs that were incurred between the date of importation and the date of resale. Under this reading, it might be difficult to conclude that a cost incurred after the date of resale was a cost incurred ‘between importation and resale’.

We are cognizant, however, that dictionary definitions can only take the interpreter so far, and that in interpreting a provision of a treaty we must take into account both context and object and purpose.(83) As discussed above, it is clear that the purpose of allowances to construct an export price is not to insure price comparability per se. Rather, an export price is constructed, and the appropriate allowances made, because it appears to the investigating authorities that the export price is unreliable because of association or a compensatory arrangement between the exporter and the importer or third party. By working backwards from the price at which the imported products are first resold to an independent buyer, it is possible to remove the unreliability. Thus, we agree with the United States that the purpose of these allowances is to construct a reliable export price to use in lieu of the actual export price or, as expressed by the EC as third party, to arrive at the price that would have been paid by the related importer had the sale been made on a commercial basis.

Read in light of this object and purpose, we recognize that costs related to the resale transaction but not incurred in a temporal sense between the date of importation and resale could as a general matter be considered to be ‘incurred between importation and resale’ and thus deducted in order to construct an export price. Nor do we preclude that an amount to cover the risk of non-payment might be considered to be such a cost. We do not believe, however, that this interpretation of costs ‘incurred between importation and resale’ can be stretched to include costs that not only were not incurred in an accounting sense until after the date of resale but which were entirely unforeseen at that time. In this regard, we observe that, while we agree with the United States that as a general principle a related importer may be expected to establish price based on costs plus profit, a price certainly cannot be expected to reflect an amount for costs that were entirely unforeseen at the time the price was set. To deduct costs which not only were incurred after the date of resale but which were entirely unforeseen at that time would not result in a ‘reliable’ export price in the sense of the price that would have been paid by the related exporter had the sale been made on a commercial basis.”(84)

(e) Fifth sentence

67. In US — Hot-Rolled Steel, the Appellate Body considered that “the obligation to ensure a ‘fair comparison’” under Article 2.4 “lies on the investigating authorities, and not the exporters. It is those authorities which, as part of their investigation, are charged with comparing normal value and export price and determining whether there is dumping of imports.”(85)

(f) Article 2.4.1

(i) Scope of Article 2.4.1

68. In US — Stainless Steel, the complainant, Korea, argued that Article 2.4.1 is the only provision of the Anti-Dumping Agreement that addresses exchange rates and the permissible modification to the dumping calculation methodology to account for exchange rate fluctuations, and thus, that the use of multiple averaging periods to account for the depreciation of the Korean won during the period of investigation was inconsistent with Article 2.4.1. The Panel responded as follows:

“In our view, Article 2.4.1 relates to the selection of exchange rates to be used where currency conversions are required. It establishes a general rule — conversion should be made using the rate of exchange on the date of sale — and an exception to this general rule for sales on forward markets. It also establishes special rules in the case of fluctuations and sustained movements in exchange rates. We note Korea’s view that the requirements of the second sentence of Article 2.4.1 prescribe specific results, rather than describing a method for selecting exchange rates. It appears to us, however, that, read in context, these special rules also relate to the selection of exchange rates, and not to the construction of averages. Rather, the permissibility of the use of multiple averaging is an issue addressed by Article 2.4.2.

Even if Article 2.4.1 were not restricted to the issue of the selection of exchange rates, we find nothing in that Article that would prohibit a Member from addressing, through multiple averaging, a situation arising from a currency depreciation. Korea contends, and the United States does not dispute, that the provision of Article 2.4.1 requiring Members to allow exporters sixty days to adjust their export prices to sustained movements in exchange rates applies only in the case of currency appreciation, and not in the case of currency depreciation. Assuming that the parties are correct in this regard, the requirement that a Member take certain actions in the case of currency appreciation does not in our view mean that Members are prohibited from taking any action to address a situation arising from a currency depreciation.(86)”(87)

(ii) “required”

69. In US — Stainless Steel, the complainant, Korea, argued that while Article 2.4.1 permits currency conversions only when such conversions are “required”, i.e., when there is no other reasonable alternative, the United States’ authority had performed an unnecessary “double conversion” of Korean local sales by converting the dollar amounts appearing in their invoices into won at one exchange rate and converting them back into dollars at a different exchange rate, in order to compare the prices of the local sales with those of exports to the United States. The Panel found that the conversions were not required because the prices being compared were in the same currency (dollars), and thus concluded that the currency conversions were inconsistent with Article 2.4.1:

“While Article 2.4.1 does not spell out the precise circumstances under which currency conversions are to be avoided, we consider that it does establish a general — and in our view, self-evident — principle that currency conversions are permitted only where they are required in order to effect a comparison between the export price and the normal value. We note that a contrary interpretation would call into doubt the utility of the introductory clause of Article 2.4.1. If the drafters had not intended to establish a rule that currency conversions be performed only when required, they could easily have drafted Article 2.4.1 to provide that ‘Currency conversions should be made using the rate of exchange on the date of sale ….’ Further, such an interpretation could result in the unusual situation where currency conversions that were required in order to perform a comparison under Article 2.4 would be subject to the rules set forth in Article 2.4.1, but unnecessary currency conversions could be performed without regard to the rules of Article 2.4.1.

We need not here arrive at any general understanding as to when currency conversions are or are not required within the meaning of Article 2.4.1, nor do we express any view regarding Korea’s ‘reasonable alternative’ test. …”(88)

70. In US — Stainless Steel, one of the issues in the context of Article 2.4.1 was whether the Korean local sales had been made for United States dollars or Korean won. The Panel stated that “if the amount of won actually paid was based on the dollar amount identified in the invoice at the market rate of exchange on the date of payment (which, because the local sales in question were letter of credit sales, came some months after the date of invoice), then the controlling amount would be the dollar amount appearing in the invoice.”(89)

(iii) Relationship with Article 2.4

71. In US — Stainless Steel, the complainant, Korea, argued that certain factual determinations of the United States’ authority on currency conversion were inconsistent with Article 2.4 as well as Article 2.4.1. The Panel held that the United States’ determination which it found consistent with Article 2.4.1 was also consistent with Article 2.4 (90), but that with respect to the other determination, which it found in violation of Article 2.4.1, “we do not consider it necessary to examine Korea’s claim that those double conversions breached a more general ‘fair comparison’ requirement under Article 2.4 of the AD Agreement.”(91)

72. In EC — Tube or Pipe Fittings, the Panel found that Article 2.4.1 “refers to currency conversion in connection with the prices of export sales, rather than to any conversion that may occur in the calculation of specific adjustments to either the normal value or the export price”.(92) It thus concluded that “the obligations concerning currency conversions in Article 2.4.1 do not apply to all conversions made in order to calculate adjustments under Article 2.4.1 — we can conceive of certain situations in which differences affecting price comparability that might lead to an adjustment under Article 2.4 might not correspond precisely with the date of the export sale (e.g. credit and warranty expenses), and where conversion of all currency data as at the date of export sale might therefore distort a fair comparison.”(93)

(g) Article 2.4.2

(i) “margins”

73. In EC — Bed Linen, the Panel interpreted the word “margins” in Article 2.4.2 as meaning the individual margin of dumping determined for each of the investigated exporters and producers of the product under investigation, for that particular product. The Appellate Body agreed with this interpretation.(94)

74. In EC — Bed Linen, the Appellate Body stated with reference to the text of Article 2.4.2, that “[f]rom the wording of this provision, it is clear to us that the Anti-Dumping Agreement concerns the dumping of a product, and that, therefore, the margins of dumping to which Article 2.4.2 refers are the margins of dumping for a product.”(95)

75. In US — Lumber V, the Appellate Body further clarified its position that “‘margins of dumping’ can be found only for the product under investigation as a whole, and cannot be found to exist for a product type, model, or category of that product”.(96) On this basis, the Appellate Body rejected the argument that zeroing would be allowed as long as all comparable transactions had been taken into consideration at the model or type level:

“It is clear that an investigating authority may undertake multiple averaging to establish margins of dumping for a product under investigation. In our view, the results of the multiple comparisons at the sub-group level are, however, not ‘margins of dumping’ within the meaning of Article 2.4.2. Rather, those results reflect only intermediate calculations made by an investigating authority in the context of establishing margins of dumping for the product under investigation. Thus, it is only on the basis of aggregating all these ‘intermediate values’ that an investigating authority can establish margins of dumping for the product under investigation as a whole.

We fail to see how an investigating authority could properly establish margins of dumping for the product under investigation as a whole without aggregating all of the ‘results’ of the multiple comparisons for all product types. There is no textual basis under Article 2.4.2 that would justify taking into account the ‘results’ of only some multiple comparisons in the process of calculating margins of dumping, while disregarding other ‘results’. If an investigating authority has chosen to undertake multiple comparisons, the investigating authority necessarily has to take into account the results of all those comparisons in order to establish margins of dumping for the product as a whole under Article 2.4.2. Thus we disagree with the United States that Article 2.4.2 does not apply to the aggregation of the results of multiple comparisons.”(97)

(ii) Weighted average normal value / weighted average export price

76. In EC — Bed Linen the Appellate Body examined the first method under Article 2.4.2 for establishing the existence of margins of dumping, i.e. the comparison of a weighted average normal value with a weighted average of prices of all comparable export transactions. The Appellate Body found the European Communities’ practice of “zeroing”(98) inconsistent with this method because by, inter alia, zeroing the negative dumping margins, the European Communities had not taken fully into account the entirety of the prices of some export transactions:

“[W]e recall that Article 2.4.2, first sentence, provides that ‘the existence of margins of dumping’ for the product under investigation shall normally be established according to one of two methods. At issue in this case is the first method set out in that provision, under which ‘the existence of margins of dumping’ must 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 …’

Under this method, the investigating authorities are required to compare the weighted average normal value with the weighted average of prices of all comparable export transactions. Here, we emphasize that Article 2.4.2 speaks of ‘all’ comparable export transactions. As explained above, when ‘zeroing’, the European Communities counted as zero the ‘dumping margins’ for those models where the ‘dumping margin’ was ‘negative’. As the Panel correctly noted, for those models, the European Communities counted ‘the weighted average export price to be equal to the weighted average normal value … despite the fact that it was, in reality, higher than the weighted average normal value.’(99) By ‘zeroing’ the ‘negative dumping margins’, the European Communities, therefore, 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. Thus, the European Communities did not establish ‘the existence of margins of dumping’ for cotton-type bed linen on the basis of a comparison of the weighted average normal value with the weighted average of prices of all comparable export transactions — that is, for all transactions involving all models or types of the product under investigation. 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.”(100)

77. In US — Lumber V, the Appellate Body confirmed its view that an authority is not allowed to practice zeroing when using the weighted-average to weighted-average comparison methodology for calculating the margin of dumping:

“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.(101) Zeroing thus inflates the margin of dumping for the product as a whole.”(102)

“comparable export transactions”

78. In EC — Bed Linen, the Appellate Body specifically addressed the term “comparable” used in Article 2.4.2, which the European Communities referred to as a basis for its appeal. More specifically, the European Communities claimed that Article 2.4.2 requires a comparison with a “weighted average of prices of all comparable export transactions” which, in the view of the European Communities, was not the same as requiring a comparison with a weighted average of all export transactions:

“In our view, the word ‘comparable’ in Article 2.4.2 does not affect, or diminish in any way, the obligation of investigating authorities to establish the existence of margins of dumping on the basis of ‘a comparison of the weighted average normal value with the weighted average of prices of all comparable export transactions’. (emphasis added)

The ordinary meaning of the word ‘comparable’ is ‘able to be compared’. ‘Comparable export transactions’ within the meaning of Article 2.4.2 are, therefore, export transactions that are able to be compared. …

…

… All types or models falling within the scope of a ‘like’ product must necessarily be ‘comparable’, and export transactions involving those types or models must therefore be considered ‘comparable export transactions’ within the meaning of Article 2.4.2.”(103)

79. In support of its proposition that the term “comparable” in Article 2.4.2 did not detract from the obligation of investigating authorities to consider all relevant transactions, the Appellate Body in EC — Bed Linen referred to Article 2.4 as part of the context of Article 2.4.2:

“Article 2.4 sets forth a general obligation to make a ‘fair comparison’ between export price and normal value. This is a general obligation that, in our view, informs all of Article 2, but applies, in particular, to Article 2.4.2 which is specifically made ‘subject to the provisions governing fair comparison in [Article 2.4]’. Moreover, Article 2.4 sets forth specific obligations to make comparisons at the same level of trade and at, as nearly as possible, the same time. Article 2.4 also requires that ‘due allowance’ be made for differences affecting ‘price comparability’. We note, in particular, that Article 2.4 requires investigating authorities to make due allowance for ‘differences in … physical characteristics’.

We note that, while the word ‘comparable’ in Article 2.4.2 relates to the comparability of export transactions, Article 2.4 deals more broadly with a ‘fair comparison’ between export price and normal value and ‘price comparability’. Nevertheless, and with this qualification in mind, we see Article 2.4 as useful context sustaining the conclusions we draw from our analysis of the word ‘comparable’ in Article 2.4.2. In our view, the word ‘comparable’ in Article 2.4.2 relates back to both the general and the specific obligations of the investigating authorities when comparing the export price with the normal value. The European Communities argues on the basis of the ‘due allowance’ required by Article 2.4 for ‘differences in physical characteristics’ that distinctions can be made among different types or models of cotton-type bed linen when determining ‘comparability’. But here again we fail to see how the European Communities can be permitted to see the physical characteristics of cotton-type bed linen in one way for one purpose and in another way for another.”(104)

Non-comparable types

80. In EC — Bed Linen, the Panel found that the European Communities “zeroing” practice was inconsistent with Article 2.4.2.(105) The European Communities appealed this finding on the ground that the word “comparable” in Article 2.4.2 indicates that, where the product under investigation consists of various “non-comparable” types or models, the investigating authorities should first calculate “margins of dumping” for each of the “non-comparable” types or models, and, then, at a subsequent stage, combine those “margins” in order to calculate an overall margin of dumping for the product under investigation. The Appellate Body disagreed with the European Communities:

“We see nothing in Article 2.4.2 or in any other provision of the Anti-Dumping Agreement that provides for the establishment of ‘the existence of margins of dumping’ for types or models of the product under investigation; to the contrary, all references to the establishment of ‘the existence of margins of dumping’ are references to the product that is subject of the investigation. Likewise, we see nothing in Article 2.4.2 to support the notion that, in an anti-dumping investigation, two different stages are envisaged or distinguished in any way by this provision of the Anti-Dumping Agreement, nor to justify the distinctions the European Communities contends can be made among types or models of the same product on the basis of these ‘two stages’. Whatever the method used to calculate the margins of dumping, in our view, these margins must be, and can only be, established for the product under investigation as a whole. We are unable to agree with the European Communities that Article 2.4.2 provides no guidance as to how to calculate an overall margin of dumping for the product under investigation.”(106)

81. In US — Lumber V, the Appellate Body clearly stated that multiple averaging, using models or types, is as such permitted under Article 2.4.2 to establish the existence of margins of dumping for the product under investigation:

“We agree with the participants in this dispute that multiple averaging is permitted under Article 2.4.2 to establish the existence of margins of dumping for the product under investigation. We disagree with those who suggest that the Appellate Body Report in EC — Bed Linen is premised on an assumption that multiple averaging is prohibited. The issue of multiple averaging was not before the Appellate Body in EC — Bed Linen and the reasoning of the Appellate Body in that case should therefore not be read as prohibiting that practice. This is not to say that EC — Bed Linen is not relevant in this appeal. Indeed, there are a number of relevant findings to which we refer to below. However, the Appellate Body did not rule on multiple averaging in that case and therefore it is incorrect to argue, as the United States does, that ‘[t]he agreement of both parties to this dispute and a unanimous Panel that Article 2.4.2 permits multiple comparisons is a fundamental departure from the premise’ of the Appellate Body Report in EC — Bed Linen.”(107)

Sampling of domestic transactions

82. The Panel on Argentina — Poultry Anti-Dumping Duties addressed the issue of whether or not a Member must include all domestic sales transactions when establishing “a weighted average normal value” for the purpose of Article 2.4.2:

“In examining what is meant by ‘a weighted average normal value’, we attach particular importance to the meaning of the term ‘normal value’. We note that Article 2.1 of the AD Agreement refers to normal value as ‘the comparable price, in the ordinary course of trade, for the like product when destined for consumption in the exporting country’. Article 2.1 therefore defines normal value in terms of domestic sales transactions in the exporting Member (although Article 2.2 provides that alternative methods to establish normal value may be used in certain circumstances).(108) Article 2.1 does not specify, however, whether or not all domestic sales transactions need be included. This issue is addressed by Article 2.2.1, which sets out the conditions to be met before domestic sales may be treated as not in ‘the ordinary course of trade’, and therefore excluded for the purpose of establishing normal value in accordance with Article 2.1. Article 2.2.1 states that domestic sales ‘may be disregarded in determining normal value only if’ the relevant conditions are met. We understand these provisions to mean that there are only specific circumstances in which domestic sales transactions may be excluded from normal value. We consider that these provisions constitute relevant context for interpreting the phrase ‘a weighted average normal value’, since they indicate that ‘a weighted average normal value’ is a weighted average of all domestic sales other than those which may be disregarded pursuant to Article 2.2.1 of the AD Agreement.”(109)

83. The Panel on Argentina — Poultry Anti-Dumping Duties thus came to the conclusion that “the strict rules in Article 2 regarding the determination of normal value require that, in the usual case, normal value should be established by reference to all domestic sales of the like product in the ordinary course of trade”.(110)

Multiple averages

84. In US — Stainless Steel, the Panel examined Korea’s argument that Article 2.4.2 prohibits the following method used by the United States authorities: (i) dividing a period of investigation into two sub-periods corresponding to the pre- and post-devaluation periods; (ii) calculating a weighted average margin of dumping for each sub-period; and (iii) combining these weighted averages of margin of dumping, however, treating sub-periods where the average export price was higher than the average normal value as sub-periods of zero dumping. In this regard, the Panel rejected Korea’s claim that Article 2.4.2 prohibits the use of multiple averaging per se :

“Article 2.4.2 provides that the existence of dumping shall normally be established ‘on the basis of a comparison of a weighted average normal value with a weighted average of all comparable export transactions’ (emphasis added). The inclusion of the word ‘comparable’ is in our view highly significant, as in its ordinary meaning it indicates that a weighted average normal value is not to be compared to a weighted average export price that includes non-comparable export transactions.(111) It flows from this conclusion that a Member is not required to compare a single weighted average normal value to a single weighted average export price in cases where certain export transactions are not comparable to transactions that represent the basis for the calculation of the normal value.

We recall Korea’s view that the reference in the singular to ‘a weighted average normal value’ means that the use of multiple averages is prohibited. In our view, however, the reference in the singular to ‘a weighted average normal value’ means simply that there must be a single weighted average normal value and export price in respect of comparable transactions. It does not mean that a Member is required to compare a single weighted average normal value to a single weighted average export price in cases where some of the export transactions are not comparable to the transactions that represent the basis for the normal value.

An examination of the context of the provision in question and of its object and purpose in our view provide further support for the above conclusion. The chapeau of Article 2.4 states that ‘[a] fair comparison shall be made between the export price and the normal value.’ Whatever the relationship of the fair comparison language of the chapeau to the specific requirements of Article 2.4 — an issue of dispute between the parties — it is evident to us that the provisions of Article 2.4.2 must be read against the background of this basic principle. In fact, the provisions of Article 2.4.2 itself are ‘subject to the provisions governing fair comparison in paragraph 4.’ An interpretation of Article 2.4.2 that required a Member to compare transactions that were not comparable would run counter to this basic principle.

Accordingly, we conclude — and by the later phases of this dispute the parties agreed — that Article 2.4.2 does not preclude the use of multiple averages per se. Rather, Article 2.4.2 requires a Member to compare a single weighted average normal value to a single weighted average export price in respect of all comparable transactions. A Member may however use multiple averages in cases where it has determined that non-comparable transactions are involved.”(112)

85. Despite rejecting Korea’s argument in US — Stainless Steel, that Article 2.4.2 precludes the use of multiple averages per se (see paragraph 84 above), the Panel found a violation of Article 2.4.2 by the United States investigating authorities. The Panel examined whether the existence of significant differences in normal value over the course of an investigation is, in and of itself, a sufficient basis to conclude — as the United States authorities had done — that export and home market transactions at different points in the period of investigation are not “comparable”:

“In examining this question, we first note that the term ‘comparable’ has been defined to mean ‘able to be compared (with)’. This definition however does not cast great light on the meaning of the term as used in Article 2 of the AD Agreement. Thus, we consider it useful to turn to the context in which this term appears. In this respect, we agree with the parties that the meaning of the term ‘comparable’ as used in Article 2.4.2 can best be established by an examination of other provisions of Article 2 of the AD Agreement that address the issue of comparability. We further note that the chapeau to Article 2.4 provides that the comparison between the export price and the normal value shall be made ‘in respect of sales made at as nearly as possible the same time’. Thus, we consider it clear that the timing of sales may have implications in respect of the comparability of export and home market transactions.(113)

This does not mean, however, that where an average to average comparison methodology is used, individual home market and export sales that are not made at the same time necessarily are not comparable and thus cannot be included in the weighted averages. To the contrary, it is in the very nature of an average to average comparison that, for example, transactions made at the beginning of the averaging period in the export market will be made at a different moment in time than sales in the home market made at the end of averaging period. If the drafters had considered that this situation would necessarily give rise to a problem of comparability, surely they would not have explicitly authorized the use of averaging in Article 2.4.2. Thus we consider that, in the context of weighted average to weighted average comparisons, the requirement that a comparison be made between sales made at as nearly as possible the same time requires as a general matter that the periods on the basis of which the weighted average normal value and the weighted average export price are calculated must be the same.”(114)

Length of averaging periods

86. The Panel on US — Stainless Steel rejected the United States’ argument that the “same time” requirement of Article 2.4 implies a preference for shorter rather than longer averaging periods, and stated:

“… If the requirement to compare sales at ‘as nearly as possible the same time’ means that sales within an averaging period covering a [period of investigation (‘POI’)] are not comparable, then a Member presumably would be obligated to break a POI into as many sub-periods as possible. Yet to interpret the word ‘comparable’, when combined with the requirement that sales be compared ‘at as nearly as possible the same time’, to obligate Members to perform numerous average to average comparisons based on the shortest possible time periods would in effect read the Article 2.4.2 authorization to perform average to average comparisons out of the AD Agreement, leaving Members with only the second option, the comparison of normal values and export prices on a transaction-by-transaction basis.(115)”(116)

87. Having found that Members are not obliged to divide a period of investigation into as many sub-periods as possible, the Panel on US — Stainless Steel nevertheless placed the following caveat:

“We do not preclude that there may be factual circumstances where the use of multiple averaging periods could be appropriate in order to insure that comparability is not affected by differences in the timing of sales within the averaging periods in the home and export markets. We note that, where changes in normal value, export price or constructed export price during the course of the POI are combined with differences in the relative weights by volume within the POI of sales in the home market as compared to the export market, the use of weighted averages for the entire POI could indicate the existence of a margin of dumping that did not reflect the situation at any given moment within the POI.(117) In this situation a Member might in our view be justified in concluding that differences in timing of sales in the home and export markets give rise to a problem of comparability that could be addressed through multiple averaging periods.(118) We recall however that this situation only arises where two elements — a change in prices and differences in the relative weights by volume within the POI of sales in the home market as compared to the export market — exist. Thus, while a change in normal value, export price or constructed export price may be a necessary condition for the conclusion that the passage of time affects comparability in the case of an average-to-average comparison, the existence of such a change is not in itself a sufficient condition to conclude that the export transactions are not comparable to the normal value.”(119)

(iii) Weighted average / individual transactions

Targeted dumping

88. The Appellate Body in EC — Bed Linen rejected the European Communities appeal that the Panel’s interpretation would not allow Members to counter dumping “targeted” to certain types of the product under investigation. With respect to the notion of “targeted” dumping, the Appellate Body referred to Article 2.4.2, second sentence, and stated:

“This provision allows Members, in structuring their anti-dumping investigations, to address three kinds of ‘targeted’ dumping, namely dumping that is targeted to certain purchasers, targeted to certain regions, or targeted to certain time periods. However, neither Article 2.4.2, second sentence, nor any other provision of the Anti-Dumping Agreement refers to dumping ‘targeted’ to certain ‘models’ or ‘types’ of the same product under investigation. It seems to us that, had the drafters of the Anti-Dumping Agreement 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. The European Communities has not demonstrated that any provision of the Agreement implies that targeted dumping may be examined in relation to specific types or models of the product under investigation. Furthermore, we are bound to add that, if the European Communities wanted to address, in particular, dumping of certain types or models of bed linen, it could have defined, or redefined, the product under investigation in a narrower way.(120)”(121)

(h) Relationship between subparagraphs of Article 2.4

89. With respect to the relationship between Article 2.4 and Article 2.4.1, see paragraph 71 above.

90. With respect to the relationship between Article 2.4 and Article 2.4.2, see paragraph 86 above.

(i) Relationship with other paragraphs of Article 2

91. With respect to the relationship between Article 2.4 and Article 2.2, see paragraph 50 above.

6. Article 2.6

92. The Panel on US — Lumber V considered that the “like product” to the product under consideration has to be determined on the basis of Article 2.6, but that this provision does not provide any guidance on the way in which the “product under investigation” is to be determined:

“Article 2.6 therefore defines the basis on which the product to be compared to the ‘product under consideration’ is to be determined, that is, a product which is either identical to the product under consideration, or in the absence of such a product, another product which has characteristics closely resembling those of the product under consideration. As the definition of ‘like product’ implies a comparison with another product, it seems clear to us that the starting point can only be the ‘other product’, being the allegedly dumped product. Therefore, once the product under consideration is defined, the ‘like product’ to the product under consideration has to be determined on the basis of Article 2.6. However, in our analysis of the AD Agreement, we could not find any guidance on the way in which the ‘product under consideration’ should be determined.”(122)

7. Relationship with other Articles

93. With respect to the relationship between Article 2 and Articles 6.1, 6.2 and 6.9, see paragraph 441 below.

94. With respect to the relationship between Article 6.8 and Articles 2.2 and 2.4, the Panel on US — Steel Plate, having found a violation of Article 6.8, considered it unnecessary to determine, in addition, whether the circumstances of that violation also constituted a violation of Article 2.4 (and Article 9.3, and Articles VI:1 and 2 of GATT 1994). In the Panel’s view, findings on these claims would serve no useful purpose, as they would neither assist the Member found to be in violation of its obligations to implement the ruling of the Panel, nor would they add to the overall understanding of the obligations found to have been violated. The Panel also declined to rule on India’s claim under Article 2.2.(123)

95. With respect to the relationship between Article 2.4 and Article 6.10, see paragraph 443 below.

96. With respect to the relationship between Articles 2.4.1 and 12, see paragraph 564 below.

8. Relationship with other WTO Agreements

(a) Article VI of the GATT 1994

97. The Panel on US — 1916 Act (EC) found that where the complainant had not established a prima facie case of violation of Article 2.1 and 2.2, “[t]he fact that we found a violation of Article VI:1 of the GATT 1994 is not as such sufficient to conclude that Articles 2.1 and 2.2 of the Anti-Dumping Agreement have been breached, in the absence of more specific arguments and evidence.”(124)

98. The Appellate Body on EC — Tube or Pipe Fittings considered that the “precise rules relating to the determination as to whether there is dumping and, if dumping exists, how the dumping margin is to be calculated, are set out, not in Article VI:2 of the GATT 1994, but rather in Article 2 of the Anti-Dumping Agreement, which is the agreement on the implementation of Article VI of the GATT 1994.” The Appellate Body in this case rejected the argument that the opening sentence of Article VI:2 of GATT 1994, “in order to offset or prevent dumping” imposed an obligation on an investigating authority to select a particular comparison methodology under Article 2.4.2 of the Anti-Dumping Agreement:

“In our view, therefore, Article 2 is a more appropriate source than the opening phrase ‘[i]n order to offset or prevent dumping’ of Article VI:2, for ascertaining specifically what is required for the proper determination of dumping by an investigating authority. We are unable to see an obligation flowing from the opening phrase of Article VI:2 of the GATT 1994 to Article 2 of the Anti-Dumping Agreement that the determination of dumping must be based on the standard of a ‘reasonable assumption for the future’, or that this, in turn, would require that a particular methodology be chosen under Article 2.4.2.”(125)

(b) Article X of the GATT 1994

99. The Panel on US — Stainless Steel touched on the relationship between Article X:3(a) of the GATT 1994 and Article 2.4.1 of the Anti-Dumping Agreement. See the Chapter on the GATT 1994, Section XI.B.D.2.

Footnotes:

1. In Marrakesh, the Ministers adopted the Decision on Anti-Circumvention, see Section XXIV. back to text
2. Appellate Body Report on US — 1916 Act, para. 119. back to text
3. Appellate Body Report on US — 1916 Act, para. 119. back to text
4. Panel Report on US — 1916 Act (EC), para. 6.208. back to text
5. Panel Report on EC — Tube or Pipe Fittings, para. 7.107. back to text
6. Panel Report on Guatemala — Cement II, para. 8.296. back to text
7. Panel Report on US — Stainless Steel, para. 6.138. back to text
8. Panel Report on US — Stainless Steel, para. 6.138. back to text
9. G/ADP/M/16, Section I, in particular, para. 84. The text of the recommendation can be found in G/ADP/6, para. 3. back to text
10. Appellate Body Report on EC — Tube or Pipe Fittings, para. 78 back to text
11. Appellate Body Report on EC — Tube or Pipe Fittings, para. 80. back to text
12. Appellate Body Report on EC — Tube or Pipe Fittings, para. 81. back to text
13. (footnote original) The United States’ perception seems to be based on the assumption that there is a watertight separation between the provision relating to construction of the export price (Article 2.3) and that relating to comparison between export price/constructed export price and normal value (Article 2.4). It is evident from the face of the text, however, that the rules regarding allowances related to construction of the export price are found in the paragraph relating to comparison. back to text
14. Panel Report on US — Stainless Steel, paras. 6.90–6.91. back to text
15. Appellate Body Report on US — Hot-Rolled Steel, para. 165. back to text
16. Appellate Body Report on US — Hot-Rolled Steel, paras. 166, 167 and 169. The Appellate Body could not, however, continue the analysis of whether the United States authorities had made any specific allowances in this case so as to make a fair comparison under Article 2.4 because it found that there was not an adequate factual record for it to complete the analysis. Para. 180. back to text
17. Appellate Body Report on US — Hot-Rolled Steel, para. 139. back to text
18. Appellate Body Report on US — Hot-Rolled Steel, para. 139. back to text
19. Appellate Body Report on US — Hot-Rolled Steel, para. 139. back to text
20. Appellate Body Report on US — Hot-Rolled Steel, para. 147. back to text
21. Appellate Body Report on US — Hot-Rolled Steel, para. 148 back to text
22. Appellate Body Report on US — Hot-Rolled Steel, para. 140. The Appellate Body also gives some examples of what it could be considered as sales not in the ordinary course of trade: “We can envisage many reasons for which transactions might not be ‘in the ordinary course of trade’. For instance, where the parties to a transaction have common ownership, although they are legally distinct persons, usual commercial principles might not be respected between them. Instead of a sale between these parties being a transfer of goods between two enterprises which are economically independent, transacted at market prices, the sale effectively involves a transfer of goods within a single economic enterprise. In that situation, there is reason to suppose that the sales price might be fixed according to criteria which are not those of the marketplace. The sales transaction might be used as a vehicle for transferring resources within the single economic enterprise. Thus, the sales price may be lower than the ‘ordinary course’ price, if the purpose is to shift resources to the buyer, who then receives goods worth more than the actual sales price. Or, conversely, the sales price may be higher than the ‘ordinary course’ price, if the purpose is to shift resources to the seller, who receives higher revenues for the sale than would be the case in the marketplace. There are many reasons relating to corporate law and strategy, and to fiscal law, which may lead to resources being allocated, in these ways, within a single economic enterprise.” Para. 141. back to text
23. Panel Report on US — Hot-Rolled Steel, para. 7.112. back to text
24. Appellate Body Report on US — Hot-Rolled Steel, para. 158. The Appellate Body also looked at another method used by the authorities, although not used in this case, which regards high-priced sales between affiliates. This so-called “aberrationally high” test excludes high-priced sales between affiliates from the calculation of normal value only if they were “aberrationally” or “artificially” high. The Appellate Body conclude that “[i]n our view, there is a lack of even-handedness in the two tests applied by the United States, in this case, to establish whether sales made to affiliates were ‘in the ordinary course of trade’. The combined application of these two rules operated systematically to raise normal value, through the automatic exclusion of marginally low-priced sales, coupled with the automatic inclusion of all high-priced sales, except those proved, upon request, to be aberrationally high priced. The application of the two tests, thereby, disadvantaged exporters.” Para. 154. As regards the Appellate Body’s conclusions as to the investigating authorities’ duty to exercise their discretion in an even-handed way, see para. 16 of this Chapter. back to text
25. Appellate Body Report on US — Hot-Rolled Steel, para. 142. back to text
26. (footnote original) One example of such a transaction is a liquidation sale by an enterprise to an independent buyer, which may not reflect “normal” commercial principles. back to text
27. Appellate Body Report on US — Hot-Rolled Steel, para. 143. back to text
28. Appellate Body Report on US — Hot-Rolled Steel, paras. 145–146. back to text
29. Panel Report on Guatemala — Cement II, para. 8.183. back to text
30. Panel Report on US — DRAMS, para. 6.66. back to text
31. Panel Report on Egypt — Steel Rebar, para. 7.393. back to text
32. Panel Report on US — Lumber V, para. 7.237. back to text
33. Appellate Body Report on US — Lumber V, para. 138. back to text
34. Appellate Body Report on EC — Hormones, para. 98. back to text
35. Panel Report on US — DRAMS, paras. 6.68–6.69. back to text
36. Panel Report on US — Lumber V, para. 7.267. back to text
37. (footnote original) The Concise Oxford Dictionary of Current English (Clarendon Press, 1995), p. 1021. back to text
38. Panel Report on US — Lumber V, para. 7.265. back to text
39. Appellate Body Report on EC — Tube or Pipe Fittings, paras. 97–98. back to text
40. Appellate Body Report on EC — Tube or Pipe Fittings, para. 101. back to text
41. Panel Report on EC — Bed Linen, para. 6.62. back to text
42. Panel Report on EC — Bed Linen, paras. 6.59–6.61. back to text
43. Panel Report on Thailand — H-Beams, para. 7.111. back to text
44. Panel Report on Thailand — H-Beams, paras. 7.112–7.113. back to text
45. Panel Report on Thailand — H-Beams, paras. 7.114–7.115. back to text
46. (footnote original) We note that in a case where there is data relating to only one other exporter or producer, a Member may have recourse to the calculation method set forth in Article 2.2.2(iii), provided, of course, that the specific requirements for the use of this calculation method are met. We recall that Article 2.2.2(iii) states that amounts for SG&A and profits may be calculated on the basis of : “any other reasonable method, provided that the amount for profit so established shall not exceed the profit normally realized by other exporters or producers on sales of products of the same general category in the domestic market of the country of origin.” back to text
47. Appellate Body Report on EC — Bed Linen, paras. 74–75. back to text
48. (footnote original) Panel Report on EC — Bed Linen, para. 6.87. back to text
49. (footnote original) It is worthwhile noting that “realized” is a word used with respect to both gains (profits) and losses. See Black’s Law Dictionary (West Group, 1999), p. 1271, which speaks of both “realized gain” and “realized loss”. back to text
50. Appellate Body Report on EC — Bed Linen, para. 80. back to text
51. Appellate Body Report on EC — Bed Linen, paras. 82–83. Following the excerpted paragraphs, the Appellate Body cited its Report on India — Patents, para. 45. back to text
52. Panel Report, EC — Bed Linen (Article 21.5 — India), para. 6.81. back to text
53. Panel Report, EC — Bed Linen (Article 21.5 — India), para. 6.84. back to text
54. Panel Report, EC — Bed Linen (Article 21.5 — India), para. 6.84. back to text
55. Panel Report on EC — Bed Linen, para. 6.94. back to text
56. Panel Report on EC — Bed Linen, paras. 6.96–6.98. back to text
57. Panel Report on Thailand — H-Beams, paras. 7.122–7.123. back to text
58. Panel Report on Thailand — H-Beams, para. 7.124. back to text
59. Panel Report on Thailand — H-Beams, paras. 7.122–7.125. back to text
60. Panel Report on Egypt — Steel Rebar, para. 7.388. back to text
61. Panel Report on EC — Tube or Pipe Fittings, para. 7.150. back to text
62. Panel Report on Egypt — Steel Rebar, para. 7.335. back to text
63. (footnote original) In this regard, we note that earlier provisions in Article 2, namely Article 2.2 including all of its sub-paragraphs, and Article 2.3, have to do exclusively and in some detail with the establishment of normal value and export price, and in addition that Article 2.1 has to do in part with the establishment of the export price. back to text
64. Panel Report on Egypt — Steel Rebar, paras. 7.333–7.334. back to text
65. Panel Report on US — Stainless Steel, para. 6.104. back to text
66. Panel Report on Argentina — Ceramic Tiles, paras. 6.113 and 6.116. A similar view was expressed by the Panel on EC — Tube or Pipe Fittings which considered that “the requirement to make due allowance for such differences, in each case on its merits, means that the authority must at least evaluate identified differences in taxation with a view to determining whether or not an adjustment is required to ensure a fair comparison between normal value and export price under Article 2.4 of the Anti-Dumping Agreement, and then to make an adjustment where it determines this to be necessary on the basis of this evaluation”. Panel Report on EC — Tube or Pipe Fittings, para. 7.157. See also Panel Report on US — Lumber V, paras. 7.165–7.167. back to text
67. Panel Report on Egypt — Steel Rebar, para. 7.352. back to text
68. Panel Report on EC — Tube or Pipe Fittings, para. 7.178. back to text
69. Appellate Body Report on US — Hot-Rolled Steel, para. 177. back to text
70. Panel Report on US — Lumber V, para. 7.165 back to text
71. Panel Report on US — Lumber V, para. 7.357. back to text
72. Panel Report on US — Stainless Steel, para. 6.75. back to text
73. (footnote original) We note however that such a situation might more properly be considered to be an “other difference… affecting price comparability”. back to text
74. Panel Report on US — Stainless Steel, paras. 6.76–6.77. back to text
75. (footnote original) Although in our view the existence of different levels of non-payment during prior periods would appear to be much more relevant. back to text
76. (footnote original) The United States contends that, “during the period of investigation, POSCO actually recognized greater bad debt expenses, as a proportion of sales, in the US market than in the Korean market. This evidence would indicate that POSCO should be charging higher prices in the US market than in the Korean market.” In the absence of any evidence in the record that the level of non-payment in the US market was foreseeable or that the historical risk of non-payment was higher in the US than the Korean market, the conclusion that POSCO should have been charging higher prices in the US than in the Korean market seems unwarranted. back to text
77. Panel Report on US — Stainless Steel, para. 6.78. back to text
78. (footnote original) But see Appellate Body Report on US — FSC, fn. 124. back to text
79. (footnote original) It can be assumed that a Member will use this authorization where appropriate without being legally constrained to do so. By contrast, the AD Agreement provides that due allowance “shall” be made for differences affecting price comparability. Mandatory language is used here because the failure to make such allowances could generate or inflate dumping margins to the detriment of the interests of other Members. back to text
80 (footnote original) That the use of the non-mandatory phrase “should” does not support the conclusion advanced by the United States can be confirmed by replacing “should” with another non-mandatory term, “may”. That a Member “may” make certain allowances would indicate that the Member was authorized but not required to make those allowances. It does not follow, however, that the Member was free also to make any other allowances not within the scope of the authorization. Cf. Appellate Body Report on US — 1916 Act, paras. 112–117 (that Article VI:2 of GATT 1994 makes imposition of antidumping duties permissive does not mean that a Member may impose measures other than anti-dumping duties to counteract dumping). back to text
81. (footnote original) As the Appellate Body stated in United States — Standards for Reformulated and Conventional Gasoline, “[a]n interpreter is not free to adopt a reading that would result in reducing whole clauses or paragraphs of a treaty to inutility.” Appellate Body Report on US — Gasoline, p. 23. back to text
82. Panel Report on US — Stainless Steel, paras. 6.93–6.95. back to text
83. (footnote original) As the Appellate Body has noted, “dictionary meanings leave many interpretive questions open.” Appellate Body Report on Canada — Aircraft, para. 153. back to text
84. Panel Report on US — Stainless Steel, paras. 6.98–6.100. back to text
85. Appellate Body Report on US — Hot-Rolled Steel, para. 178. back to text
86. (footnote original) The provision relied upon by Korea is the language in Article 2.4.1 stating that, “in an investigation the authorities shall allow exporters at least 60 days to have adjusted their export prices to reflect sustained movements in exchange rates during the period of investigation”. Korea is in effect asking us to read this provision to further say that “in an investigation the authorities shall take no actions to address currency depreciations”. We can perceive no textual basis to imply such an additional rule into Article 2.4.1. back to text
87. Panel Report on US — Stainless Steel, paras. 6.129–6.130. back to text
88. Panel Report on US — Stainless Steel, paras. 6.11–6.12. back to text
89. Panel Report on US — Stainless Steel, para. 6.25. However, pursuant to Article 17.6(i), the Panel did not find one factual determination of the US authority on this issue in violation of Article 2.4.1. See para. 636 of this Chapter. back to text
90. Panel Report on US — Stainless Steel, para. 6.44. back to text
91. Panel Report on US — Stainless Steel, para. 6.45. back to text
92. Panel Report on EC — Tube or Pipe Fittings, para. 7.198. back to text
93. Panel Report on EC — Tube or Pipe Fittings, para. 7.199. back to text
94. Panel Report on EC — Bed Linen, para. 6.118. Appellate Body Report on EC — Bed Linen, para. 53. This interpretation was also confirmed by the Appellate Body in US — Hot-Rolled Steel, para. 118. back to text
95. (footnote original) Appellate Body Report on EC — Bed Linen, para. 51. back to text
96. Appellate Body Report, US — Lumber V, para. 96 back to text
97. Appellate Body Report, US — Lumber V, paras. 97 – 98. back to text
98. The European Communities practice of “zeroing” was summarized in the Panel Report on EC — Bed Linen as follows: first, the European Communities identified with respect to the product under investigation — cotton-type bed linen — a certain number of different “models” or “types” of that product. Next, the European Communities calculated, for each of these models, a weighted average normal value and a weighted average export price. Then, the European Communities compared the weighted average normal value with the weighted average export price for each model. For some models, normal value was higher than export price; by subtracting export price from normal value for these models, the European Communities established a “positive dumping margin” for each model. For other models, normal value was lower than export price; by subtracting export price from normal value for these other models, the European Communities established a “negative dumping margin” for each model. For these latter models, in other words, dumping had not occurred, as the export price exceeded the normal value. The European Communities then calculated the overall dumping margin by averaging the individually calculated results for the different models, but counting “negative dumping margins” as zero in the process. The Panel found that this practice was inconsistent with Article 2.4.2. Panel Report on EC — Bed Linen, para. 7.1(g). back to text
99. (footnote original) Panel Report on EC — Bed Linen, para. 6.115. back to text
100. Appellate Body Report on EC — Bed Linen, paras. 54–55. back to text
101. (footnote original) We note that the Panel reached the same conclusion in para. 7.216 of its Report. back to texf
102. Appellate Body Report on US — Lumber V, para. 101. back to text
103. Appellate Body Report on EC — Bed Linen, paras. 56–58. back to text
104. Appellate Body Report on EC — Bed Linen, paras. 59–60. back to text
105. Panel Report on EC — Bed Linen, para. 7.1(g). For the description of the zeroing practice, see footnote 98. back to text
106. Appellate Body Report on EC — Bed Linen, para. 53. back to text
107. Appellate Body Report on US — Lumber V, para. 81. back to text
108. (footnote original) These methods are not relevant in the present proceedings, since the DCD established normal value on the basis of domestic sales transactions. back to text
109. Panel Report on Argentina — Poultry Anti-Dumping Duties, para.7.272. back to text
110. Panel Report on Argentina — Poultry Anti-Dumping Duties, para.7.274. back to text
111. (footnote original) We note that insertion of the word “comparable” into Article 2.4.2 represented the only modification to that Article between the date of the Draft Final Act and the text as adopted. See Draft Final Act Embodying the Results of the Uruguay Round of Multilateral Trade Negotiations, MTN.TNC/W/FA, 20 December 1991. This suggests that its inclusion was not merely incidental but reflected careful consideration by the drafters. back to text
112. Panel Report on US — Stainless Steel, paras. 6.111–6.114. back to text
113. (footnote original) As an additional contextual argument, Korea argues that devaluation cannot be considered to affect comparability because there is no provision in the AD Agreement specifying that sales made at one exchange rate cannot be compared with sales at another exchange rate. Rather, the only provision of the AD Agreement that addresses exchange rates is Article 2.4.1, which the United States concedes does not establish a limit on what sales may be considered comparable. We do not however place any weight on Korea’s argument in this respect. In our view — and absent the unusual situation of multiple exchange rates — there will at any given moment in time be only one exchange rate. Thus, any problem of comparability does not relate to exchange rates per se, but rather to differences in timing of sales. Thus it is on this issue that we focus. back to text
114. Panel Report on US — Stainless Steel, paras. 6.120–6.121. back to text
115. (footnote original) The United States’ argument seems to be posited on its view that the best comparison for measuring dumping is a transaction-to-transaction comparison, and that average-to-average comparisons are a second-best approach allowed because of practical problems with the transaction-to-transaction methodology. See US answer to question 2 from the Panel posed at the second meeting of the Panel with the parties. We perceive no valid textual basis for such a conclusion, however. To the contrary, the AD Agreement sets forth two options for a comparison methodology — average-to-average and transaction-to transaction — and expresses no preference between them. back to text
116. Panel Report on US — Stainless Steel, para. 6.122. back to text
117. (footnote original) A particularly dramatic example of this situation would arise where, during a substantial portion of the POI, there were no sales in one of the two markets. back to text
118. (footnote original) The combination of these two factors could even result in a situation where, although at any given moment in time throughout the POI, the exporter was charging an identical price (after all appropriate allowances had been made), a margin of dumping could nevertheless be found to exist. For example, imagine that there were two home market sales (HM-1 and HM-2) and two export sales (EX-1 and EX-2) during the POI. HM-1 and EX-1 occurred on day 1 and were both at a price of $10. HM-2 and EX-2 occurred on day 90 and were both at a price of $15. Thus, neither of the individual export transactions was dumped when compared to the simultaneous home market transactions. If all these sales were in the same volumes, then a weighted average to weighted average would also show no dumping. Assume however that HM-1 and EX-2 involved a volume of ten units, while HM-2 and EX-1 involved a volume of twenty units. In this case, the weighted average normal value would be (10 units × $10/unit) + (20 units × $15/unit) = $400/30 units = $13.33/ unit. The weighted average export price would be (20 units × $10/unit) + (10 units × $15/unit) = $350/30 units = $11.27/ unit. Thus, the weighted average margin of dumping would be 18 per cent. back to text
119. Panel Report on US — Stainless Steel, para. 6.123. back to text
120. (footnote original) The European Communities also argues in its appellant’s submission, paras. 42–45, that the Panel’s interpretation of Article 2.4.2 would disadvantage those importing Members which collect anti-dumping duties on a “prospective” basis when compared to those importing Members which collect anti-dumping duties on a “retrospective” basis. We note, though, that Article 2.4.2 is not concerned with the collection of anti-dumping duties, but rather with the determination of “the existence of margins of dumping”. Rules relating to the “prospective” and “retrospective” collection of anti-dumping duties are set forth in Article 9 of the Anti-Dumping Agreement. The European Communities has not shown how and to what extent these rules on the “prospective” and “retrospective” collection of anti-dumping duties bear on the issue of the establishment of “the existence of dumping margins” under Article 2.4.2. back to text
121. Appellate Body Report on EC — Bed Linen, para. 62. back to text
122. Panel Report on US — Lumber V, para. 7.153. back to text
123. Panel Report on US — Steel Plate, para. 7.103. back to text
124. Panel Report on US — 1916 Act (EC), para. 6.209. back to text
125. Appellate Body Report on EC — Tube or Pipe Fittings, para. 76. back to text