1. Introduction
This paper analyzes the decision by the Article 22.6 DSU arbitrator in US–Anti-Dumping Methodologies (China).Footnote 1 We recall that the Panel and Appellate Body (AB) reports in the underlying dispute US–Anti-Dumping Methodologies (China) were adopted by the DSB in May 2017.Footnote 2 The AB had found that the United States’ use of the weighted average-to-transaction (WA-to-T) methodology combined with simple zeroing as well as the treatment of various Chinese exporters as a single ‘China Inc.’ entity under the Single Rate Presumption (SRP)Footnote 3 had violated various provisions of the WTO Anti-Dumping Agreement (ADA).Footnote 4
An Article 21.3(c) DSU arbitrator had granted the US a reasonable period of time (RPT) of essentially seven months to implement the AB findings (until 22 August 2018), but when the US failed to do so, China requested the DSB authorization to suspend concessions vis-à-vis the US worth US$ 7.043 billion. The US objected and asked for Article 22.6 DSU arbitration. The arbitrator decided that China was entitled to suspend concessions in the amount of US$ 3,579.128 million, i.e. roughly half of what China had requested.Footnote 5
While this may seem like a huge amount compared to, for example, the US$ 74.40 million granted to Korea in US–Washing Machines (Article 22.6 – US),Footnote 6 it is important to bear in mind that US–Anti-Dumping Methodologies (China) covered 25 anti-dumping orders.
2. Scope of the Dispute
In the original dispute, the Panel had found both ‘as such’ and ‘as applied’ violations.
The ‘as applied’ violations on the one hand concerned violations of Article 2.4.2 ADA in three original investigationsFootnote 7 where the US had used the WA-to-T comparison method with simple zeroing and a violation of Article 9.3 ADA in an administrative review investigationFootnote 8 where the US had applied this method. On the other hand, they concerned violations of Articles 6.10 and 9.2 ADA in 13 original investigationsFootnote 9 and 25 administrative reviews where the US Department of Commerce (USDOC) had applied a single anti-dumping duty to Chinese companies that did not meet its Separate Rate Test (SRT).
In addition, the Panel had found that the USDOC's SRP constituted a measure of general and prospective application that was ‘as such’ inconsistent with Articles 6.10 and 9.2 ADA.Footnote 10
In its estimation of the level of nullification or impairment, China provided calculations for the ‘as applied’ violations with regard to the 13 original investigations above. China further claimed that the ‘as such’ violations affected more than a hundred anti-dumping duty orders but limited its calculations to 12 additional anti-dumping duty orders.Footnote 11
Even though China thus provided calculations for 25 orders, its claim of US$ 7.043 billion was based only on the ‘as applied’ violations in the 13 original investigations. However, China requested that the Arbitrator include the level of nullification or impairment resulting from the 12 ‘as such’ violation findings if the latter were to reject or lower China's calculations regarding the ‘as applied’ violations. The Arbitrator agreed with this approach.Footnote 12
China later chose to exclude one of the ‘as such’ cases, namely Aluminum Extrusions,Footnote 13 unless the Arbitrator were to use another calculation methodology other than that used by China. Despite objections from the US, the Arbitrator agreed with this approach.
The US requested exclusion of the OTR Tires Footnote 14 case on the grounds that the USDOC had revoked the order. However, China objected on the grounds that the order had still been in force for up to six months after the expiry of the RPT. The majority of the Arbitrators agreed with China.Footnote 15
3. The Counterfactual
One of the most interesting aspects of the case concerns the Arbitrator's selection of the appropriate counterfactual. As the Arbitrator noted:
Counterfactuals are frequently used by arbitrators and reflect “a hypothetical scenario that describes what would have happened in terms of trade flows had the responding party implemented the DSB recommendations and rulings”
… [W]e will evaluate whether China's proposed counterfactual reflects “at least a plausible or ‘reasonable’ compliance scenario ...
… If we find that this counterfactual does not reflect a reasonable or plausible compliance scenario, we will proceed to determine an alternative counterfactual … Footnote 16
China claimed that the ‘as applied’ violations found with respect to the 13 original investigations tainted the entire orders. Therefore, they argued, the counterfactual should be based on the withdrawal of the anti-dumping duties imposed on all Chinese exporters under these orders.
The Arbitrator did not agree with this ‘all or nothing’ approach but took the more nuanced view that only certain elements of the orders had been found to be WTO inconsistent. Thus, neither injury/causation aspects nor procedural elements of the orders were affected by the violation findings. Furthermore, different Chinese exporters were impacted to a different extent by the violations. For example, the duties imposed with regard to most sampled exporters (with the exception of those for which zeroing had been applied) and with regard to the non-sampled exporters that had met the SRT were not affected by the violations found.
In the view of the Arbitrator, the counterfactual should reflect what would have happened if the US, after the RPT expiry, had ceased using the WA-to-T comparison with zeroing as well as the SRP, even if this was more complicated and required a case-by-case analysis.Footnote 17 This was essentially the counterfactual approach proposed by the United States.Footnote 18
3.1 Weighted Average-to-Transaction with Zeroing
In the Coated Paper Footnote 19 case, the USDOC had made calculations for the sampled exporter APP China both by using the WA-to-WA method and by using the WA-to-T method with zeroing. Under the former, the dumping margin for APP China was de minimis while under the latter it amounted to 7.62%. The 7.62% rate had also been used for the non-sampled Chinese exporters that passed the SRT. The US argued, and the Arbitrator agreed, that for these two groups of exporters a zero duty should be used as the counterfactual. A similar approach was taken in Oil Country Tubular Goods Footnote 20, it being understood that the dumping margin for the sampled producer Tianjin Pipe Corporation (TPCO) was substantially higher under both the WA-to-WA and the WA-to-T with zeroing methods.
In Steel Cylinders Footnote 21, the duty for the sampled producer Beijing Tianhai Industry Co., Ltd. (BTIC) using the WA-to-T method with zeroing amounted to 6.62%. While that duty had been revoked, the same duty continued to be applied to non-sampled Chinese producers that had met the SRT. As there was no WA-to-WA calculation on the record available, the Arbitrator decided that a zero-duty rate by analogy to the Coated Paper approach represented a reasonable and plausible counterfactual.
Finally, in PET Film Footnote 22 the USDOC in the original investigation had calculated a duty rate of 3.49% for the sampled Chinese exporter DuPont by using the WA-to-T with zeroing method. However, in the subsequent fourth administrative review, non-sampled DuPont had been assigned a separate duty rate of 31.24%, based on the WTO consistently calculated duty rate of two other sampled Chinese exporters. The United States argued, and the Arbitrator agreed, that it had ceased its use of the illegal WA-to-T with zeroing method prior to the expiry of the RPT as far as DuPont was concerned and that, therefore, there was no nullification or impairment.
3.2 Single Rate Presumption
We recall that in many anti-dumping cases, the authorities establish three types of duties, namely (1) individual duties for sampled producers, (2) a duty for non-sampled cooperating producers, based on the (weighted average) dumping margin found with respect to the sampled producers,Footnote 23 and (3) a residual or ‘all others’ duty for non-sampled, non-cooperating producers which tends to be relatively high as it often is based on adverse facts available (AFA).
The USDOC followed a different approach when applying duties on China. Chinese exporters that met the SRT and were not subjected to AFA would fall in the second category while Chinese exporters that failed the SRT would typically fall in the third category, even though some of them would have been considered as cooperating if they had met the SRT test. However, on the basis of the single economic entity theory underlying the SRP concept, the failure of even one company in the ‘China Inc.’ group to cooperate would be treated as a failure to cooperate of all the companies in the group.Footnote 24
Table 1 shows the differences between the two approaches.
Standard approach vs. Single Rate Presumption Approach
The United States argued that within the third group a distinction should be made between cooperating and non-cooperating producers and that only for the cooperating producers a counterfactual, based on the separate duty rates imposed in the investigations, needed to be used; the non-cooperating producers in any event would have been subjected to the People's Republic of China-wide (PRC-wide) duty rate, calculated on the basis of AFA.Footnote 25
China disagreed with the US position on two grounds. First, it considered that within the group of allegedly non-cooperating producers a distinction should be made between producers that considered that they had in fact provided all the necessary information to the USDOC and that, in any event, the facts available used should have been ‘neutral’ and not ‘adverse.’ Second, China argued that for the cooperating producers in the third group the separate duty rates did not constitute an appropriate counterfactual because such a counterfactual assumed that the separate rates had been calculated in a WTO consistent manner, which was not the case. In the view of China, WTO inconsistencies included the USDOC's illegal use of AFA, double-counting duties from anti-dumping and countervailing duty investigations,Footnote 26 as well as its differential pricing methodology and the use of the WA-to-T with zeroing comparison method.
The Arbitrator rightly identified that China's argumentation raised the issue of whether an Arbitrator in an Article 22.6 DSU dispute can take into account possible WTO inconsistencies that were not raised in the original dispute and (a majority) decided that it could.
The Arbitrator considered that while it was not within its mandate to make findings with respect to the WTO consistency of a measure, ‘it would be incongruous to assess whether a counterfactual reflects a reasonable or plausible compliance scenario without considering that counterfactual's WTO consistency.’Footnote 27 It therefore decided to include in its assessment not only the WTO violations that had been established in the original proceeding, but also violations of other WTO obligations.
The US acknowledged that in the original and administrative review investigations in dispute, it had not made a distinction within the third group between cooperating and non-cooperating producers, but rather had found that the PRC-wide entity as such had failed to cooperate. Taking into account that the USDOC thus had not reached company-specific findings of non-cooperation and did not follow the Article 6.8 and Annex II ADA procedures for reaching facts available findings, the Arbitrator considered that the continued use of the PRC-wide duty rate would not constitute a reasonable or plausible compliance scenario for any of the producers in the third group.Footnote 28 Thus, the Arbitrator rejected the distinction proposed by the United States.
The Arbitrator further considered that use of the separate duty rates, as proposed by the US for the cooperating producers in the third group, would not be an appropriate counterfactual in light of the right provided in Article 6.10.2 ADA for non-sampled producers to request individual examination and the obligation in that same article for the administering authorities to conduct such an examination unless it would be unduly burdensome and prevent the timely completion of the investigation. The Arbitrator did not consider it reasonable to assume that none of the producers within the PRC-wide group would have requested individual examination or that the USDOC would have been permitted to reject all such requests because they would have been unduly burdensome.Footnote 29
The Arbitrator also considered it relevant that the separate duty rates had not in all cases been set in accordance with Article 9.4 ADA because the USDOC, in response to questions by the Arbitrator, had acknowledged that such rates in certain cases had included duties based on AFA and in other cases had been based on the duty rates alleged by the petitioner in the complaint or on duty rates calculated in previous proceedings.Footnote 30
As the USDOC had not calculated alternative duty rates for the producers within the PRC-wide group, the Arbitrator considered it reasonable to apply, ‘as a proxy’, a zero-duty rate as the counterfactual for such producers.Footnote 31
In a separate opinion, one member considered that the Arbitrator should have only considered the violations found in the original Panel dispute settlement proceeding and, consequently, should have used the separate rate counterfactual only for the cooperating producers in the third group.
4. The Quantification of the Nullification or Impairment: Quantifying the Damage
4.1 The Task of the Arbitrator
The task of the Arbitrator was to determine the level of nullification or impairment of benefits that accrued to China as a result of the US violation or breach found by the Dispute Settlement Body. In other words, the panel set out to determine what US imports from China would have been (the ‘but for’ levels) had it not been for the WTO inconsistent aspects of the US antidumping duty measures following the expiration of the RPT. The damages would then be equivalent to the difference between actual and ‘but for’ levels. Damage calculations rely heavily on how the counterfactual is defined; that is, what would have been the case had it not been for the inconsistent measures?
With 25 antidumping duty determinations, three methodologies to consider (weighted average-to-transaction, single rate presumption, and zeroing), and across various products and firms, quantifying the extent to which subject trade flows were hindered from only the inconsistent duties was not a trivial task. Further complicating the task was the Arbitrator's view that only certain elements of the orders had been found to be WTO inconsistent.
The United States and China each proposed a methodology. Each party noted the objective using the withdrawal of the inconsistent measures as the counterfactual, but the analytical approaches differ, as we briefly note below. The Arbitrator rejected both parties’ proposals and decided to utilize a two-step approach to estimate damages. We describe the two-step approach below and the strengths and weaknesses of the Arbitrator's approach.
4.2 The Methodology Proposed by the United States
The US proposed a basic partial equilibrium Armington-based approach, using data from 2017 and estimating the trade effects of removing the violation. Interested observers will recognize this as an Armington-based imperfect substitute trade model, often referred to as the so-called Compas model (Francois and Hall, 2003).Footnote 32 Armington models have been applied widely for the analysis of trade policy changes despite the strong consequences of assumptions on key elasticities.
4.3 The Methodology Proposed by China
China countered that the duties depressed trade and hence using 2017 would have meant a smaller initial market share than China would had but for the inconsistent duty. Instead, China proposed a difference-in-difference tabular approach to estimate the impact of the AD order by comparing the following two scenarios: (i) the evolution of US imports from China subject to the anti-dumping order, defined as ‘the treatment group’, between the period prior to the imposition of the order and the year 2017; and, (ii) the evolution of US imports from a group of countries not subject to anti-dumping orders on the product at issue, defined as ‘the comparison group.’Footnote 33
4.4 Arbitrator's Methodology
The Arbitrator rejected both of the proposed approaches and instead came up with its own two-step approach. The first step is a partial equilibrium analysis of the trade effect of the initial antidumping duty imposed by Commerce that incorporates the WTO inconsistent portion. The second step is a similar partial equilibrium approach but uses the resulting market share from the first step as the starting point and then estimates the trade effect of removing the inconsistent portion of the duty using data from the RPT year. The nullification and impairment is the difference between the two estimated values.
The following simple example with made-up values elucidates key aspects of the Arbitrator's special methodology:
Suppose the US imports blue widgets from China, and further suppose that the US imposes 50% anti-dumping duties on these subject goods from China. After some time, China brings the case to the WTO and the panel finds that some of the AD duties (say, half) were inconsistent with WTO commitments.
Under the two-step approach:
• First, the panel runs the PE model and finds the trade effect of the 50% anti-dumping duty was a decline in China's market share from 40% to 15%.
• Second, the panel estimates the trade effects of removing the inconsistent portion of the duty but starts with the estimated 15% market share from the first step. (This second step uses all three estimated market shares: US, China, and rest of world.). That is, suppose China has 15% of the market, and then the duty decreased from 50% to 25%. The PE model will show an increase in China's market share as a result of the lower duty. Suppose the results indicate China's market share increases from 15% to 25%.
• The impairment is the difference between the 15% and 25% market share. That is, had it not been for the inconsistent portion of the duties, China's market share would have been 10 percentage points greater. (We note that the nullification or impairment should be the difference in revenue flowing to China but in this approach the estimated market shares are inclusive of the tariff.) Damages are calculated as 10% of the market value of the subject good in the remedy year.
4.5 Note on Panel's Two Step Approach
There are two key problems with the two-step approach. First, the approach ignores the effects of other market factors during the period such as changes in supply, demand, or consumer preferences. The longer the period, the more likely there would be structural changes in those factors. Of the 25 cases under consideration, some overlapped with major macroeconomic events that affected trade patterns such as supply chain shifts that were well underway before the duty was imposed.
Second, the approach ignores co-existing duties not in violation that were imposed at the same time or after the inconsistent duties were imposed. Such duties would affect market shares and those effects would not be captured in this two-step approach.
Third, if the original duties were prohibitive and only a portion of the prohibitive duties were inconsistent, the results might not be plausible. For instance, suppose the original duties were 100%, and the market share decreased significantly. Then suppose the Arbitrator found some of those duties to be WTO inconsistent. The first step would show a large decrease in market share (because of the effect of the prohibitive duties even though only some were WTO inconsistent), but then a too-small market share would be used in the second step, and Armington-based models are notorious for getting stuck in the corner.
A cleaner approach may have been to simply assess the trade effect of imposing the full duties and only the WTO consistent duties and taking the difference in the resulting market shares (using the year of the initial duty). The market value to which that change in market share is applied however is tricky. Perhaps a transparent approach could be a simple annual average over the period. Without doing a simple average or another transparent approach, the Arbitrator would be left to disentangle relevant and irrelevant market forces. That is, the more time that has passed since the initial imposition of the tariffs, the greater likelihood there were other market forces driving change, with some of those market forces directly or indirectly related to the tariff, and others unrelated.
Notwithstanding these three economic analysis critiques, the panel was given a nearly impossible task of disentangling the effects of a myriad of measures over numerous cases and time periods. Single rate presumption alone is extremely difficult to quantify because it requires the identification of the firms affected and not affected by the SRP, and the corresponding years that each firm was and was not affected by the SRP. Also, to disentangle which elements of the order were WTO inconsistent would require determining the alternative AD rate ‘but for’ the SRP.
On one hand, by assuming inconsistencies with WTO obligations other than those that were found to have been violated in the original proceedings, it seems the Arbitrator may have gone above and beyond its mandate. On the other hand, the alternatives were bleak. Indeed, in such complex cases with multiple duties (some deemed WTO consistent others not), spanning multiple firms, across multiple products, and over various time periods, perhaps we are asking too much of the Arbitrators to identify a reasonable counterfactual.
5. Assessment
WTO dispute settlement cases involving trade remedy measures tend to be different from cases involving other areas of WTO law because the underlying measures are highly fact specific. Thus, for example, a calculation of an anti-dumping duty consists of many variables each of which could theoretically be challenged by the exporting country government. To give an extreme example, suppose that an exporter has made 5,000 export transactions, that the administering authority has wrongly deducted too much ocean freight for one transaction and refuses to correct it. The exporting country government takes the case to the WTO and wins. If the dispute goes all the way to Article 22.6 DSU arbitration, then what should be the level of nullification or impairment? Should it be the difference between the correctly calculated anti-dumping duty and the erroneously calculated duty, or should it be assumed that the whole duty/measure was illegally imposed?
WTO Arbitrators have taken the position that it should only be the difference between the two,Footnote 34 at least in cases where both calculations form part of the administrative record, as was the case here for Coated Paper and Oil Country Tubular Goods. It would seem that in the similar Steel Cylinders case where the WA-to-WA calculation was not part of the record, the Arbitrator gave the affected Chinese producers the benefit of the doubt, as the calculated duty rate (6.62%) on the basis of the WA-to-T method with zeroing was similar to the one found under that method in Coated Paper (7.62%). It seems to us that this was a reasonable extrapolation on the basis of the facts of the case.
Our assessment, however, is different with respect to the counterfactual used by the Arbitrator with regard to the SRP violation, i.e., a zero duty for all Chinese producers covered by the PRC-wide duty (rather than the separate rate duty imposed with respect to the non-sampled cooperating producers that met the SRT). Different from the Steel Cylinders case, the separate rate duties appear to have been available and therefore could have been used. However, starting from its position that it can take into account (possible) violations of WTO rules other than those found in the original dispute, the Arbitrator found a violation in that some of the producers in the PRC-wide group, if they had not have been included in that group, might have requested individual examination and that the USDOC might have decided to grant such requests. But thus far in practice in the US and other countries, requests for individual examination, especially by Chinese exporters, are relatively rare and the granting of such requests is even more exceptional. Therefore, to use this possible violation as a justification to use a zero-duty counterfactual in all the cases is extremely far-fetched.
Overall, the complexity of this case is highly concerning. The two-step approach was earlier utilized in a far simpler case, US–Washing Machines,Footnote 35 which involved just one investigation, not 25, and there were no arguments about the correct data to use. Further, the calculation of the nullification or impairment of the violation was relatively straightforward (e.g., zeroing), and the duty rates were fairly low. But the two-step approach may not be applicable to other cases, especially those which are overly complex such as this one.
The second possible violation is the incorrectly calculated separate duty rates in some cases, which seems to have been acknowledged by the United States. Effectively assuming, however, that such a violation occurred in all 25 cases seems a bridge too far. On the other hand, for the Arbitrator to examine in which of the 25 cases the separate rate duty was incorrectly calculated also seems unreasonable – such a task would require the WTO staff to have complete data from the original investigation, the computer code, and WTO staff to make all the adjustments to the pricing data. Further, the WTO staff would need to be able to identify the set of firms in the first year of the duty subject to the SRP, the duty those firms would have had without the SRP, the group of firms subject to the SRP in the remedy year and the relevant margin. Even if all such data and necessary information were made available to WTO staff, the data would be unverifiable.
Thus, in cases with multiple orders across multiple firms, multiple products, multiple years, preclusive duties, and various violations under consideration, the parties must allow the Arbitrator to use its discretion.
In other words, while a case can be made against the WTO's presumption of ‘full removal’ the alternative is far less desirable if not outright preposterous. And therein lies the problem with these bunched-up cases. At some point, the extent of the incorrect calculation becomes impractical to disentangle and the parties leave it to the discretion of the Arbitrator.
Thus, the two justifications used in this case by the Arbitrator to come up with the zero-duty counterfactual illustrate the danger of an Arbitrator taking into account (possible) violations that may have occurred in the administrative investigations, but that were never claimed by the Applicant in the original dispute settlement proceeding.
6. Conclusions
The decision of the Arbitrator in this case opens the door for re-litigating a case by including violations in the context of arguing a reasonable counterfactual that were never claimed in the original investigation and therefore never properly litigated. The two (possible) violations used by the Arbitrator in this case as the justification for using a zero duty as the counterfactual for the Chinese producers in the ‘China Inc.’ group illustrate the danger of such an approach. Thus, most administering authorities would never have granted individual examination to non-sampled exporters to any significant degree. While the USDOC may have incorrectly calculated the separate rate duty in some investigations, it seems unlikely that it would have done so in each and every one of the 25 investigations that formed the basis for the Arbitrator's decision.
Acknowledgements
The authors would like to thank Tom Prusa, Juhi Dion Sud, Chad Bown, Isabelle Van Damme and participants at the WTO Case Law Project Conference 2020 for helpful comments on previous drafts. Any remaining errors are our own.
