Skip to main content Accessibility help
Hostname: page-component-5c569c448b-qj5tk Total loading time: 0.535 Render date: 2022-07-01T17:00:43.470Z Has data issue: true Feature Flags: { "shouldUseShareProductTool": true, "shouldUseHypothesis": true, "isUnsiloEnabled": true, "useRatesEcommerce": false, "useNewApi": true } hasContentIssue true

Mexico – Antidumping Measures on Rice

‘Don't ask me no questions and I won't tell you no lies’

Published online by Cambridge University Press:  14 April 2008

Edwin B. Parker Professor of Law at Columbia Law School, Professor at the University of Neuchâtel, Research Fellow at CEPR
Professor of Economics, ECARES, Université Libre de Bruxelles, Senior Fellow at Bruegel, Research Fellow at CEPR
Rights & Permissions[Opens in a new window]


Review Article
Copyright © Petros C. Mavroidis and André Sapir 2008

1. Facts

This case (Mexico – Antidumping Measures on Rice)Footnote 1 concerns the manner in which dumping margins should be calculated for exporters who have not been individually examined during the investigation process. Mexico imposed antidumping (AD) measures on US exporters of beef and rice to its market. It did not take any particular steps in order to identify all exporters; it limited itself to an investigation of US exporters identified in the petition, as well as of US exporters who voluntarily presented themselves to the Mexican authority. Mexico imposed the highest individually calculated margin on all ‘unknown’Footnote 2 US exporters, that is on all US producers who had not been investigated and exported beef and rice to the Mexican market in the years following the imposition of antidumping duties. The United States (US) protested. In its view, Mexico should have applied the residual rate (that is, a duty corresponding to the average individual dumping margin found, as opposed to the highest individual dumping margin found) to all noninvestigated exporters, and not the highest duty, as it did.Footnote 3

The Panel and the Appellate Body (AB) did not explicitly disagree with the United States, but did not uphold its arguments either. The Panel found that the Mexican authority should have pursued the investigation more actively, whereas the AB found against Mexico because it used wrong data.

The Panel and the AB also entertained claims on the manner in which Mexico chose the period of investigation (POI). Although this was not the core issue of the dispute, we consider it necessary to discuss the relevant findings since this is an area where case law is still evolving and the legislative guidance is quite open-ended. Mexico arbitrarily (that is, without any justification) restricted the months in each calendar year during which it investigated the existence of dumping and injury; it also used data that were 15 months old. Both the Panel and the AB condemned the use of POI chosen by Mexico.Footnote 4

In Section 2, we discuss the claim concerning the duty that unidentified exporters should be paying. Section 3 deals with the claims on POI. Section 4 introduces economic considerations, and Section 5 concludes with a specific recommendation to improve the process of AD investigation.

2. Who are the dumpers?

2.1 A brief discussion of the legislative framework

Before, we examine the case, an introduction to the issue is necessary for two reasons:

  1. (a) the facts are highly complicated;

  2. (b) the AD Agreement accepts on the one hand that dumping is private business practice, but allows countrywide AD orders;

  3. (c) at the same time, the AD Agreement does not adequately define who should be investigated when countrywide AD orders are imposed. State practice has evolved in a nonlinear manner in this respect.Footnote 5

Since dumping is private practice, an investigating authority will have to identify a source of supply and calculate an individual dumping margin. Because, on the other hand, there is at least an implicit suspicion that all other private entities originating in the same country might be dumping as well, an investigating authority might be calculating dumping margins for more than one exporter. If there are many sources of supply identified, the investigation could be hampered in case each one of them was to be investigated: in such cases, an investigating authority can investigate only a sample, and apply a residual duty on all noninvestigated exporters (following the disciplines enshrined in Arts. 6.10 and 9.4 AD). Finally, new exporters (considered shipments), that is producers who originate in the country being investigated and who did not export during the investigation process, will undergo an expedited review when they start exporting to the country that has AD duties in place against imports from the country at hand.

Assuming producers have been sampled, the AD Agreement suggests that only two rates of AD duties are permissible:

  1. (a) an individually calculated duty for all sampled exporters and all other exporters who have come forward and requested to be investigated;

  2. (b) the weighted average of the individually calculated duties will be applied to all other sources (originating in the same country).

In case no sampling took place, individual dumping margins will be imposed on all investigated exporters. The AD Agreement does not explicitly allow authorities to investigate only some exporters in a nonsampling scenario. This is, nevertheless, what happened in the present case: Mexico did not sample; it calculated dumping margins for those investigated, and then imposed the highest margin on all noninvestigated exporters.Footnote 6 The Agreement does not explicitly regulate this transaction. Besides investigated exporters, the AD Agreement regulates the dumping margins to be imposed on one more category of exporters: new shipments (Art. 9.5 AD). With respect to new shipments, as briefly alluded to above, the AD Agreement requests that an individually calculated duty be imposed, but is silent on the amount of default duty (assuming one is WTO consistent), that is the duty that they have to pay until a review has been initiated. There are good arguments to support the thesis that new shipments have to pay a duty until their review has been initiated. The obligation included in Art. 9.5 AD (to start an expedited review of the duty imposed on products coming from new exporters in order to calculate an individual duty for such new exporters) implies as much. The context of Art. 9.5 AD makes this conclusion inescapable. Art. 9.5 AD talks of products on which duties have been imposed, and the need for expedited review on new shipments. If no duties were in place, why should the review be expedited? Actually if no duties were in place on new shipments and the review was expedited, new shipments would have been enjoying a less favorable regime than old shipments that profited from a normal investigation. This could raise MFN issues. It is legally impossible that the law is illegal, and the task of the interpreter is to interpret it in a manner that all provisions retain their scope. It seems to us that to do that, we must presume that the new shipments are burdened by duties before the expedited review takes place. Then comes the question of the duty to be applied to unidentified exporters, assuming for the time being that such exporters, who were exporting during the period of investigation, can exist. It seems that, in principle, the only existing possibility,Footnote 7 assuming a willingness to impose duties, is that such exporters come under the purview of Art. 9.5 AD. There is an inconvenience, however: the wording of Art. 9.5 AD seems to suggest that it covers only cases of exporters who were not exporting the product at hand during the period of investigation. Hence, an extension to cover producers that were exporting such product is arguably contra legem.

It could be, of course, that some exporters originating in the same country (with the exporters being investigated) are not known to the authority. They could be unknown for a variety of reasons: because they managed to hide (let us call this, uncooperative behavior), or because the authority did not take any reasonable efforts to identify them (for example, they continued to export and were never requested to appear before the authority), or for other reasons as well. With respect to such unknown exporters, that is exporters who were exporting to the country investigating at the time the investigation takes place but for whatever reason were not identified during the investigation process, the AD Agreement is silent as to how much duty they should be paying, assuming they should be paying at all. We know, nevertheless, that such exporters are liable to pay duties since countrywide orders are perfectly GATT consistent: WTO Members can lawfully impose duties on all producers originating in the country of the dumper (Art. 9.2 AD).

Two questions thus emerge:

  1. (a) First, what is the extent of the obligation imposed on the investigating authority to identify exporters?

  2. (b) Second, the related question, assuming that there is no obligation to identify all current exporters (that is, with the exception of new shipments), what is the amount of duty to be imposed on unknown exporters?

2.2 Speculating about the responses

Art. 6.10 AD, we recall, requests determination of individual dumping margins for all known exporters. The same provision allows for sampling of exporters, in case their number is large. The nonsampled exporters will pay a duty in accordance with Art. 9.4 AD (weighted average). The term known exporters is nowhere defined in the AD Agreement. Let us take the sampling scenario first: one way to understand this provision is that sampling will occur only after all exporters have been identified. Assuming this is the correct interpretation, there will be no room for unknown exporters, but only for new shipments (i.e., producers who did not export during the POI). The upshot of this approach is the elimination of the third category (unknown exporters). The downside is that this might incite strategic behavior on behalf of the investigating authority and might push up AD duties: an investigating authority will have the incentive to investigate individually those dumpers that, in its view, might be credited with the largest dumping margin and thus impose a high weighted average on the rest. Is there insurance policy against this risk? Yes, in principle: nonsampled producers who believe they dump less, or do not dump a lot, will come forward and ask to be individually investigated. The investigating authority can, nevertheless, thwart such requests pointing to the nonpracticability of a marginal investigation (Art. 6.10 AD).

Art. 6.10 AD could also be understood as meaning that, following a first review of the file, it appears that there are dozens of exporters. The investigating authority does not go all the way and identify all exporters. It does investigate a sample of those identified (although, not a sample of all exporters), and imposes a weighted average on all other exporters, identified or not. Here, there is room for unknown exporters, but it does not matter. Known nonsampled and unknown exporters alike, will pay the weighted average duty, as per Art. 9.4 AD.

A variant of this latter approach would distinguish between known nonsampled and unknown exporters: the former would pay the weighted average, and the latter something else. This is the EC approach, which is not challenged in the present dispute.

Does the Agreement privilege one over the other approaches? The Agreement is unclear. There are good arguments though, supporting the view that the negotiators had in mind the first and/or the second scenario, but not the third: Art. 6.10 AD states that a statistically viable sample must be investigated. This term has a particular meaning in statistical analysis and would require an active research by the investigating authority. The same provision, however, waters down this concept by adding that the sample should be statistically viable on the basis of information with the investigating authority. This last sentence could be a simple truism (how could it be otherwise? The investigating authority can judge only upon information it possesses), or could be understood as diluting the obligation imposed: it is not a statistically viable sample that must be investigated, but a statistically viable sample of the exporters known to the investigating authority. Depending on the number of known exporters, the number could be quite low. By the same token, Art. 6.10 AD requests from authorities to investigate the largest percentage of the volume of exports that can reasonably be investigated. Criteria such as administrative capacity could be quite relevant in quantifying this obligation.

And what about cases, like the one before us, where no sampling took place? Once again, an investigating authority must impose individually calculated dumping margins on known exporters (Art. 6.10 AD). Does this provision also entail that it must identify all exporters? There are good arguments in favor of this thesis. One could plausibly, for example, argue that the sampling scenario should be reserved for cases where there are dozens of exporters, and the nonsampling for cases where the number of exporters is limited. In that latter case, the investigating authority should be responsible for calculating individual margins. The counterargument here would be: what if an exporter is hiding during the investigation process? How credible is this counterargument? Recall that those hiding will pay the residual duty anyway in the US, and a higher duty in other markets (such as Mexico, or the EC). So, they would have little incentive to hide in the first place.Footnote 8 And how credible is it that they can hide in a scenario where only few exporters exist? The importing state could check the relevant data, and identify the sources of supply to its market one by one. This is more or less what the Panel, as we will see, requested from the Mexican authority to do.

To recap: sampling should be reserved to cases where it is not manageable to investigate individually all exporters; in other cases, individual margins must be calculated for all. Hence, if at all, unknown exporters will exist only in case sampling has taken place. There should be no room for unknown exporters in the nonsampling scenario: those that hide and do not export during the POI will be treated as new shipments; those that continue exporting but do not present themselves to the investigating authority will be detected through cooperation between the customs authority and the investigating authority. If their number is not large, they will be individually investigated; in the opposite scenario, the investigating authority can perfectly well proceed and sample.Footnote 9

Let us see now how all these issues were addressed by the AB in the report under discussion.

2.3 The AB approach

In the past, the AB has, on a number of occasions, revealed a preference in favor of an active investigating authority. In Mexico – Antidumping Measures on Rice, the Mexican investigating authority had limited its investigation to two US exporters identified by the petitioner, and two that had motu proprio presented themselves to the Mexican authority. The Panel held the view that the Mexican investigating authority violated its obligations by not making a reasonable effort to identify exporters other than the four mentioned above. In the case at hand, the Panel felt that one could reasonably expect that it is incumbent upon an active investigating authority to look for some easily available information.Footnote 10

The AB disagreed, but probably ended up with a similar, if not more ambitious, result. For a start, it held, following a completely textual interpretation of Art. 6.10 AD, that the term known exporters appearing in Art. 6.10 AD does not include exporters that the investigating authority should have known when the investigation was initiated (§255). As a result, the term known exporters is limited to exporters identified by the petitioners and exporters who voluntarily identified themselves to the investigating authority. By inference, unknown exporters are not only those covered by Art. 9.5 AD (that is, new shipments: exporters who were not exporting during the investigation), but also those who were exporting during the investigation but were not identified (either by the petitioners or voluntarily).

The next question before the AB was whether the AD Agreement imposes any limits on the amount of the duty to be paid by unknown exporters (who are not considered new shipments). Recall that the AD Agreement is silent on this score. US practiceFootnote 11 suggests that, indeed, the residual rate (weighted average) of the individually established dumping margins should be applied to new shipments, as well as other unknown exporters. The Panel was not convinced by this argument. In §7.159 it developed its rationale for rejecting this argument. We quote from §7.159:

The US argument that the placement of this provision immediately preceding Article 9.5 of the AD Agreement dealing with new shipper reviews implies that its rules also apply to non-shipping exporters is not convincing, as we do not find that anything can be deduced in and of itself from the sequence of provisions in the Agreement, particularly when the provision in question relates to an exceptional situation, while the subsequent provision does not. The United States also argues that the non-sampled interested parties and the new shippers dealt with by Article 9.5 are in a similar position and that by analogy the same Article 9.4 methodology for the calculation of a residual duty rate should apply. We are not convinced that the text of the Agreement supports this view. In this respect, we find particularly relevant, the absence of any cross-referencing in Article 9.5 of the AD Agreement dealing with new shippers to the calculation methodology of Article 9.4 of the AD Agreement. This absence of cross-referencing is particularly conspicuous if one were to accept, arguendo, the analogous situation of non-sampled and non-shipping exporters. Indeed, especially in such a situation, one would expect the drafters to have explicitly referred to Article 9.4 of the AD Agreement. As on other occasions, where the drafters intended to see obligations apply in similar circumstances, they explicitly provided for such cross-referencing. We recall in this respect that the Appellate Body also found that the absence of such cross-referencing to obligations contained in other provisions is revealing of the absence of such an obligation. We find that Article 9.4 of the AD Agreement does not refer to non-shipping exporters outside a sampling situation, and that there was therefore no obligation for the Mexican authorities to calculate a residual duty margin for Producers Rice based on the ‘neutral’ methodology set forth in Article 9.4 of the AD Agreement. We therefore reject the US claim in this respect. (Italics in the original).

The above-cited passage suggests that an investigating authority, when imposing duties on shipments coming from new exporters does not have to apply a duty equivalent to the weighted average. This does not mean that, if it does so, it will be violating the AD Agreement. The question, nevertheless, remains what is the maximum permissible duty under the circumstances?Footnote 12

In Mexico – Antidumping Measures on Rice, the AB did not take a clear stance on the issue. Indirectly, however, it provided some answers. The AB did state that an authority is not permitted to impose a residual duty rate based on facts available. According to the AB, an authority that imposes a duty on unidentified exporters based on facts available, including facts from the petition, is acting in violation of Art. 6.8 and paragraph 1 of Annex II AD. We quote from §§259–260:

The second sentence of paragraph 1 of Annex II conditions the use of facts from the petitioner's application on making the interested party ‘aware’ that, if the information is not supplied by it within a reasonable time, the investigating authority will be free to resort to these facts available. In other words, an exporter shall be given the opportunity to provide the information required by the investigating authority before the latter resorts to facts available that can be adverse to the exporter's interests. An exporter that is unknown to the investigating authority – and, therefore, is not notified of the information required to be submitted to the investigating authority – is denied such an opportunity. Accordingly, an investigating authority that uses the facts available in the application for the initiation of the investigation against an exporter that was not given notice of the information the investigating authority requires, acts in a manner inconsistent with paragraph 1 of Annex II to the Anti-Dumping Agreement and, therefore, with Article 6.8 of that Agreement.

The United States exporters that Economía did not investigate were not notified of the information it required. Notwithstanding this, Economía used facts available contained in the application submitted by the petitioner against these uninvestigated exporters.

According to the AB, putting exporters on notice that facts available will be used is a precondition for the use of facts available. This condition, for obvious reasons, can never be met in the case of unidentified exporters: one cannot notify the person that has not been identified.

The AB addressed a situation in which the residual rate was based on petitioner data, and was particularly adverse when compared to the margins of dumping calculated for the examined (known) exporters. But the need to inform exporters of the fact that, in the absence of cooperation, facts available will be used, applies in all cases, and not only when the data used are provided by the petitioner. To a certain extent, any margin based on information other than data provided by the exporter itself is based on ‘facts available’. Logically speaking, the AB's statement could thus be read to imply that no residual duty can be imposed on such unknown exporters. A rate based on the highest margin of an exporter individually calculated, as is the EC's practice, is also a facts-available rate for the noninvestigated exporter, and hence, in this line of thinking, would, as well, be WTO inconsistent. If this reading is correct, the AB effectively closed the door to the possibility that unknown exporters exist when no sampling has taken place: since recourse to Art. 6.8 AD is impermissible in such cases, the investigating authority must identify all nonsampled exporters.Footnote 13

To conclude on this score, the AB responded in the following manner:

  1. (a) there is no obligation for the investigating authority, in a nonsampling scenario, to identify all exporters;

  2. (b) an investigating authority cannot use facts available against nonidentified exporters;

  3. (c) it follows that an investigating authority must calculate individual dumping margins in this respect;

  4. (d) in this line of logic, the maximum duty for nonidentified exporters will be the duty that will be individually calculated.

So let us take a real-world example and see how the AB finding holds. In the instant case, Mexico identified only four exporters and calculated dumping margins for each one of them. Days or weeks later, a fifth exporter, who was exporting during the POI, exports to Mexico. Mexico must calculate a dumping margin for this exporter, since it cannot use facts available (data from the four exporters already investigated). Since this exporter is not considered a new shipment, Mexico cannot conduct an expedited review either. So, in this reading, Mexico must pay the price (assuming there is a price to pay) for not having actively investigated the transaction before it, and identified all exporters. Recall that nothing stops Mexico from sampling, in case it realizes during the investigation process that the number of exporters is quite large.

On the other hand, one could make the argument that all the AB wanted to say is that facts available (Art. 6.8 AD) will not be used against nonidentified exporters in a nonsampling scenario. What else, however, could be used? How could the Mexican authority impose duties on nonidentified exporters that it does not individually investigate without having recourse to the information it gathered by investigating the US exporters for which an individual dumping margin had been calculated? All the Mexican authority knows is the amount of dumping margin for those individually calculated. If it cannot use it through Art. 6.8 AD on nonidentified exporters, the only provision it might have recourse to is Art. 9.5 AD. Is that what the AD had in mind? Is the AB's view that nonidentified exporters should be considered new shipments? If yes, then the AB has effectively extended the coverage of this provision to transactions not envisaged by the legislator: recall that this provision covers cases of producers who were not exporting during the investigation period, and not cases of producers who were exporting but were not identified during the investigation process.

To conclude: the AB does not request from the Mexican authority, in a nonsampling scenario, to actively investigate all exporters. It does not allow it to use facts available against nonidentified exporters either. The only reasonable conclusion under the circumstances is that, if Mexico does not ask the right questions, it will have to pay the price.

3. The period of investigation (POI)

3.1 The issues raised

The complainant raised two issues:

  1. (a) it complained that Mexico chose only the six months every year (in the three-year investigation period), where import penetration was high. It thus ignored the other six months every year where the domestic industry fared better. As a result, the comparison was unfair. Mexico replied that it chose the same months for dumping and injury investigations, otherwise the comparison would be unfair;

  2. (b) it further complained that Mexico used 15-month-old data. Such data were inappropriate in light of the changes that occurred ever since. Mexico replied that these are the best data it could have used under the circumstances.

3.2 The legislative framework

The AD Agreement does not regulate the length of the POI. It imposes a requirement for fair comparison (Art. 2.4 AD), but falls short of imposing specific obligations as to the choice of POI. State practice has moved in to fill this gap. State practice, however, is not completely uniform, although elements of overlap are quite substantial. Over the years, WTO Members have, in the context of the Committee on Antidumping Practices (ADP), managed to reach some common understanding on the requirements for an appropriate choice of POI. However, two points are appropriate in this context: first, the ADP recommendations do not constitute an exhaustive treatment of the POI, since many questions are still open; second, the legal value of an ADP recommendation is an open issue.Footnote 14

The Panel Report in part relies on the ADP Committee recommendation on the length of POI to support its view as to the length of the injury POI (§7.62).Footnote 15 The use of ADP recommendations in this limited manner proved useful. The ADP recommendation, however, was of no use when the Panel and the AB tried to address the two questions asked by the complainant.

3.3 Some months every year …

Recall that the United States claimed that the AD Agreement had been violated because the Mexican authority had analyzed only six months' worth of data for each of the three years of data collection. Mexico asserted that it was necessary to examine these particular six months of every year instead of the full year in order to ensure that the period of the injury analysis paralleled the six-month period chosen for the analysis of dumping, so as to avoid any distortions.

The Panel saw no a priori reason why a period of investigation on the injury analysis should be chosen to fit the period of investigation for the dumping analysis in case the latter period of investigation covers a period of less than 12 months, as there is nothing in the AD Agreement that would require such an approach; quite to the contrary, in its view, was true.Footnote 16 The Panel considered that the choice of the period of investigation is crucial, as it determines the data that will form the basis for the assessment of the impact of dumping, and that an examination or investigation can only be ‘objective’ if it is based on data that provide an accurate and unbiased picture of what it is that one is examining.Footnote 17

Mexico chose a three-year POI, and restricted its investigation into the six months for each of the three years. In fact, Mexico upheld the POI proposed by the petitioner (interested, of course, in seeing AD duties imposed). In the Panel's view, for Mexico to legitimately restrict the POI in this manner, it would have to provide some justification why this was necessary in order to perform a fair comparison. Mexico did not do that. In fact, the only element before the Panel was the US claim that the six months chosen were the months of high import penetration. In the absence of justification for the restriction, the Panel condemned the Mexican choice in the following manner:

In sum, we find that the injury analysis of the Mexican investigating authority in the rice investigation which was based on data covering only six months of each of the three years examined, is inconsistent with Article 3.1 of the AD Agreement as it is not based on positive evidence and does not allow for an objective examination, as it necessarily, and without any proper justification, provides only a part of the picture of the situation. In addition, we find that the particular choice of the limited period of investigation in this case was not that of an unbiased and objective investigating authority as the authority was aware of, and accepted, the fact that the period chosen reflected the highest import penetration, thus ignoring data from a period in which it can be expected that the domestic industry was faring better.Footnote 18

The AB agreed that these two factors – the selective use of the information gathered and the fact that the authority accepted the POI proposed by the petitioner, knowing that the petitioner proposed that period because it allegedly represented the period of highest import penetration – plus the absence of any justification explaining why the use of data from the remaining six months would distort the picture, were sufficient to conclude that the data used by the Mexican authority did not provide an accurate and unbiased picture (§§181ff.).

It is difficult to find fault with the AB on this score. The data used seem to suffer from selection bias (assuming that it is factually correct that the six months chosen are those of the highest import penetration, a fact that Mexico has not disputed). Such data cannot provide the input for a proper evaluation of the dumping margin and the ensuing injury.

3.4 How old can data be?

The United States further claimed that the Mexican authority had used a POI for injury that ended more than 15 months prior to the initiation of the investigation. The Panel considered that while the AD Agreement does not contain any specific and express rules concerning the period to be used for data collection in an antidumping investigation, this does not mean that the authorities' discretion in using a certain period of investigation is boundless.Footnote 19 The Panel was of the view that there is necessarily an inherent real-time link between the investigation leading to the imposition of measures and the data on which the investigation is based. In spite of the fact that an antidumping investigation out of necessity relies on historical data gathered during a past POI, such information should be the most recent information reasonably available:

Of course, it is well established that the data on the basis of which this determination is made may be based on a past period, known as the period of investigation. Nevertheless, because this ‘historical’ data is being used to draw conclusions about the current situation, it follows that the more recent data is likely to be inherently more relevant and thus especially important to the investigation. This, as a consequence, implies that the data considered concerning dumping, injury and the causal link should include, to the extent possible, the most recent information, taking into account the inevitable delay caused by the need for an investigation, as well as any practical problems of data collection in any particular case.Footnote 20

This led the Panel to the following conclusion:

The requirement of a time-consuming and sometimes complicated investigation to demonstrate the existence of dumping and the ensuing injury poses a practical impediment to a complete identity in time between the imposition of the measure and the conditions for such imposition, i.e. dumping causing injury. Although this practical problem may lead to the situation in which any determination of dumping causing injury has by the time of the imposition of the measure become more of a proxy than a real time assessment of the current situation, it would, in our view, not be correct to be led by the practical necessity to examine the past to assess the present to accept that an investigating authority could justifiably base itself on old data to the exclusion of more recent data which was available and usable. To the contrary, the fact that an investigation of up to 12 months may have to be conducted to determine dumping, injury and the causal link magnifies the importance of having a period of data collection which ends as closely as possible to the date of initiation, as by the time of the possible imposition of the measure another 12 months may have passed.Footnote 21

The Panel thus considered that a 15-month gap between the end of the period of investigation and the initiation of the investigation is sufficiently long as to impugn the reliability of the period of investigation to deliver, for the purposes of a determination, evidence that has the requisite pertinence or relevance, thereby failing to meet the criterion of ‘positive evidence’ pursuant to Article 3.1 AD.Footnote 22 The Panel did not state that the AD Agreement imposed a requirement not to use 15-month-old data. It did interpret the Agreement, however, to impose a requirement to at least explain why, in case such data have been used, it still adequately explains the state of affairs when the imposition of AD duties is being decided. Mexico failed to do that.

Note that, in order to reach its conclusion, the Panel once again relied on the ADP recommendation on POI which, in pertinent part, states that that the period of data collection should end as close to the date of initiation of the investigation as is practicable.Footnote 23

The AB fully upheld the reasoning of the Panel.Footnote 24 It emphasized the fact that the determination of whether injury exists should be based on data that provide indications of the situation prevailing when the investigation takes place, because the conditions to impose an antidumping duty are to be assessed with respect to the current situation. Mexico's failure to explain why historical data could still appropriately explain the state of affairs when duties were imposed was critical in the AB's finding.Footnote 25

Here too, it is difficult to find fault with the attitude of the AB. Even in the absence of clear language to this effect, it is illogical to interpret the AD as allowing for imposition of duties based on historical data. The wording of Art. 11.1 AD is highly pertinent here: duties shall be in place as long as necessary to counteract injury caused by dumping. A necessary precondition is that injury is being caused at the moment when duties are being imposed. Knowledge that injury is being caused as a result of dumping can only be based on data that prove that this is indeed the case. There should be a presumption that this is indeed the case when the data used are as close as possible to the date when the imposition of duties is being decided. Older data could also suffice assuming that the investigating authority can prove that nothing has changed ever since.

The AB did not go all the way and establish a reversal of the presumption in case older data have been used. It would be hard to implement such a proposal as a rule of thumb since it is far from being an easy exercise to come up with an age for permissible and impermissible data that will make sense across transactions. Implicitly however, this is what its ruling amounts to. In fact, all the United States proved in this context was that the data were 15 months old. It sufficed for the Panel and the AB to reverse the burden of proof and request from Mexico to respond why such data were still appropriate.

4. The problem lies with the law, not the judges' ruling

With respect to the POI, it is difficult to find fault with the rulings by the AB. We will thus focus on the treatment of unidentified exporters. The AB seems to have limited the risk of abusing the investigation process by outlawing recourse to facts available in a nonsampling scenario. The AB, nevertheless, cannot undo the intrinsic inadequacies of the AD Agreement. The AB can, at best, try to rationalize aspects of the AD Agreement, while maintaining the balance of rights and obligations struck by the principals themselves, namely the WTO Members. In fact, however, the problem with the AD Agreement goes much deeper than the lack of clarity with respect to the treatment of unidentified exporters. It relates to whether it is countries or firms that are responsible for dumping and that should pay AD duties if they engage in injurious dumping. In what follows, we advance some thoughts on this issue, and argue that a fundamental incoherence must be removed from the AD Agreement, whereby it considers dumping as private practice on the one hand, and presumes that dumping is a countrywide practice on the other. If this incoherence remains, we risk seeing many other dumping cases being brought before the WTO adjudicating bodies. Our arguments, therefore, go beyond the present case by suggesting what we consider as a necessary legislative amendment to the AD Agreement.

Dumping is a pricing strategy whereby producers in one country charge lower prices in foreign markets than in their home market. As such it is practiced by firms, not countries. Yet Article 2 AD defines dumping in terms of countries with no reference to individual firms:

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.Footnote 26

This definition would seem to imply that the pricing strategy of all exporting firms of a particular product is identical, with the same price differential between the home market and the importing country for all relevant firms. This, obviously, makes little or no economic sense. There is no a priori reason to believe that two (or more) firms behave identically in terms of their pricing in different markets simply because they happen to produce the same product in the same country – unless they operate under perfect competition in the home- and in the importing-country markets.

If perfect competition prevailed, it would be hard to understand how dumping could occur altogether. Since dumping is a form of price discrimination, it requires that firms be able to set prices in the different markets rather than take them as given, as it is the case under perfect competition.

If markets were perfectly competitive, the ‘law of one price’ would prevail and all identical products would have only one price – at least within the same geographical or national market. However, perfect competition is clearly an extreme case that is unlikely to prevail in many circumstances, and therefore ‘the “law of one price” is no law at all’ in fact.Footnote 27

There are essentially two reasons why the ‘law of one price’ does not hold. The trivial reason is that products may not be identical either because they are differentiated or because differences in the services offered by competing firms might lead them to charge different prices for the same product. But even if products are truly homogenous, price dispersion is likely to be the rule rather than the exception to the ‘law of one price’. As Nobel Laureate George Stigler argued more than 40 years ago, price dispersion is related to the existence of imperfect information and search costs of consumers.Footnote 28

Product differentiation and imperfect information imply that firms might be able to exert some degree of market power and hence to price discriminate. Moreover, different firms can be expected to face different circumstances and to have different capabilities that result in different costs and therefore different prices – unless they operate under perfect competition in which case prices have to be the same for all firms and the different costs simply translate into different profits.

The AD Agreement actually recognizes in two places that firms which produce the same product in the same country do not necessarily behave identically. The first acknowledgment of this comes in Art. 5 AD on antidumping initiations and investigations, where the Agreement states that applications to national authorities by the domestic industry in the importing country for antidumping measures must contain:

a complete description of the allegedly dumped product, the names of the country or countries of origin or export in question, the identity of each known exporter or foreign producer and a list of known persons importing the product in question.'Footnote 29

Furthermore, Art. 6 AD on evidence in antidumping investigations states that:

The authorities shall, as a rule, determine an individual margin of dumping for each known exporter or producer concerned of the product under investigation.Footnote 30

In other words, the Agreement makes it clear that each ‘known exporter or foreign producer’ must be identified and that the national authorities must determine individual dumping margins for each of them. There is no question of lumping them all together and assuming that they behave identically.

However, as already discussed supra, there are two possible interpretations of the expression ‘known exporter or foreign producer’. One interpretation is to consider that only those exporters or producers from the country concerned who have been identified by the domestic industry in its application for AD measures have potentially engaged in injurious dumping. The other interpretation is to treat all exporters or producers from that country as if they potentially engaged in injurious dumping, and to treat those identified by the domestic industry in its application as a special subset of exporters or producers.

Art. 6 AD contains no indication as to which of the two interpretations is favored by the Agreement. However, Art. 9 AD on the imposition and collection of AD duties contains some language dealing with another category of ‘unknown’ exporters, those who were not exporting to the importing country during the period of investigation but did so later on, which provides the necessary clue:

If a product is subject to anti-dumping duties in an importing Member, the authorities shall promptly carry out a review for the purpose of determining individual margins of dumping for any exporters or producers in the exporting country in question who have not exported the product to the importing Member during the period of investigation, provided that these exporters or producers can show that they are not related to any of the exporters or producers in the exporting country who are subject to the anti-dumping duties on the product. Such a review shall be initiated and carried out on an accelerated basis, compared to normal duty assessment and review proceedings in the importing Member.Footnote 31

Hence any exporter or producer of a product and from a country subject to AD duties is to be viewed as potentially engaging in injurious dumping in the importing country, if it exported to the importing country after the period of investigation. A fortiori the same must hold for exporters or producers who exported to the importing country during the period of investigation.

The interpretation chosen by the Agreement implies that all firms producing the allegedly dumped product in the country in question engage in injurious dumping, unless proven otherwise. It puts, therefore, the burden of proving innocence on all exporters, both those who are ‘known’ and those who are ‘unknown’.

It could perhaps be argued that this interpretation of the Agreement rests not only on Art. 9.5 AD, but also on the language in Art. 6 AD itself.

The text of the Agreement states that:

All interested parties in an anti-dumping investigation shall be given notice of the information which the authorities require and ample opportunity to present in writing all evidence which they consider relevant in respect of the investigation in question.Footnote 32

The Agreement also states that interested parties include:

(i) an exporter or foreign producer or the importer of a product subject to investigation, or a trade or business association a majority of the members of which are producers, exporters or importers of such product.Footnote 33

These two statements could be interpreted as saying that national authorities in the importing country must investigate all exporters or producers of the allegedly dumped product from the exporting country subject to investigation. This would imply that, indeed, they are all considered potentially guilty of injurious dumping. Such reading of Art. 6 AD is reinforced by the statement that:

In cases in which any interested party refuses access to, or otherwise does not provide, necessary information within a reasonable period or significantly impedes the investigation, preliminary and final determinations, affirmative or negative, may be made on the basis of the facts available. The provisions of Annex II shall be observed in the application of this paragraph.Footnote 34

In other words, all interested parties must participate in the investigation, and those who refuse may be subject to AD duties.

Whatever the exact legal basis for the interpretation chosen by the Agreement, the fact of the matter is that it runs counter to economic rationality. The correct economic interpretation would have been that firms producing the same product in the same country will typically behave differently because they face different economic conditions and because perfect competition does not hold. Such interpretation would have implied that investigating authorities should not presume that all firms engage in injurious dumping simply because some might or even do.

This leads us to conclude that the Agreement is internally inconsistent from an economic viewpoint. On the one hand, the Agreement supposes that all exporters or producers of an allegedly dumped product who produce in the country in question are potentially guilty of injurious dumping in the importing country. On the other, it recognizes that different exporters or producers may behave differently and should therefore be subject to different treatment in terms of AD duties.

Hence, with respect to the fundamental question of the burden of proof and, ultimately, whether it is countries or firms that should be held responsible for dumping, nothing short of a revision of the AD Agreement itself can be satisfactory from an economic viewpoint.

5. Lock the case law

The AB advanced an interpretation of the AD Agreement that seems to move dumping back to the realm of private practices, as it should be, and away from unprovable conspiracy scenarios on which the Agreement itself in part rests. By outlawing recourse to Art. 6.8 AD in the case of nonsampling, the AB effectively imposes a duty on national investigating authorities to establish individual dumping margins for all exporters, instead of presuming that they all dump in the same manner simply because of their common origin. This is one (small) step forward, and it would be desirable, indeed, to lock it somehow. Although the lawmaking function of the AB is highly disputed, it is probably advisable to proceed first in the context of the ADP Committee and secure a RecommendationFootnote 35 that would repeat the AB findings and clarify that investigating authorities:

  1. (i) are under the duty to identify all exporters in a nonsampling scenario;

  2. (ii) must calculate individual dumping margins for each exporter.

By the same token it would be helpful to also clarify issues (legitimately) not touched upon by the present case, such as:

  1. (i) the fact that nonsampled exporters are not liable to pay duties until a review takes place;

  2. (ii) the nature of such a review (whether it should be expedited or not);

  3. (iii) the amount of duty to be paid by new shipments.

The fundamental problem that we have identified in this paper is the unjustified presumption that all producers originating in a country must be dumping because one of them does; it must be taken care of through legislative amendment.


1 WTO Doc. WT/DS295.

2 We will use the terms unknown and unidentified as synonyms throughout this paper.

3 The US government was de facto defending its own practice.

4 The AB entertained a third claim on the permissibility of using assumptions during an investigation. The claim was upheld since Mexico was found to be violating the requirements of Arts. 3.1 and 3.2 AD to base its findings on positive evidence, on procedural grounds. This was a side issue briefly discussed and better explained in other case law. This is why we decided not to discuss it in detail. For the record, however, we mention that the AB acknowledged that, within the bounds of this discretion, an investigating authority may be expected to rely on reasonable assumptions or draw inferences. In doing that, it must ensure that its determinations are based on ‘positive evidence’. Although it is not impermissible to use assumptions, these assumptions should be derived as reasonable inferences from a credible basis of facts, and should be sufficiently explained so that their objectivity and credibility can be verified. WTO members (through their investigating authorities) who use a methodology premised on unsubstantiated assumptions are ipso facto violating Art. 3.1 AD. An assumption is not properly substantiated when the investigating authority has not explained why it would be appropriate to use it in the analysis. Mexico, in the Final Determination, did not explain why the assumptions on which the investigation relied upon were appropriate and credible in the analysis of the volume and price effects of the dumped imports, or how they would contribute to providing an accurate picture of the volume and price effects of the dumped imports (§§204–205).

5 As we will argue later in this Section, this is one case where there is need for legislative amendment in order to avoid similar problems in the future.

6 Hence, the US claim that Mexico should have imposed the weighted average on noninvestigated exporters has no basis in the AD Agreement. There is an obligation to impose the weighted average on noninvestigated exporters only in case an authority has sampled exporters.

7 As we argued before, Art. 9.4 AD cannot be functionally put into operation, absent knowledge of all exporters.

8 Only assuming that they dump more than other exporters, and they know that this is indeed the case, they would still have this incentive to hide. But, even in this case, they would have no incentive to hide if they export to Mexico or the EC.

9 Recall that for this approach to be workable, cooperation among two governmental functions, customs authorities and AD investigating authority, is required. This might prove a daunting task for some developing countries. Note also, however, that the size of this problem should not be overstated: first, many developing countries, active users of the system, like Mexico, for example, have efficient customs authorities. Second, all that is required to communicate from one agency to another is information about the origin of the goods, a rather trivial issue.

10 US exporters were being identified in a commercial publication that was before the Mexican investigating authority.

11 It bears repetition that the US scheme is contra legem: the US authority applies the residual duty even in case of nonsampling, when the AD Agreement clearly reserves the possibility to apply a residual duty for cases where sampling has taken place. EC practice, as briefly alluded to above, suggests another, residual, rate that applies to imports from unknown or new exporters: exporters that are not new shippers and kept quiet during the investigation, or exporters that remained unknown during the investigation, the reasonable efforts of the EC authority to identify them notwithstanding, will see their exports burdened not with the weighted average, but with a residual rate. Assume that the EC authority has sampled three exporters who ship equal volumes to the EC market, and that they are found to be dumping by 10%, 20%, and 30% respectively. The EC authority will impose the duties mentioned above to the three investigated exporters; a 20% (weighted average) duty on all identified exporters; and a 30% (residual) rate on nonidentified exporters as well as on new shipments. It seems that the EC higher duty for unknown exporters is reserved for cases when sampling has taken place. The consistency of such practice with the multilateral rules has not been established as yet.

12 The Panel, in its report on Mexico – Antidumping Measures on Rice, did not have to address this issue.

13 A very imaginative reading of the AB Report could lead to the conclusion that all the AB wanted to say was that facts available which are adverse may not be used to calculate a duty for unknown exporters. There remains a serious problem: when is the use of facts available (not) adverse? Is a residual duty based on the highest margin found for an investigated exporter less adverse than the use of information contained in the petition? Is a lesser duty (for example, weighted average) not so adverse? If this was the true intention of the AB, it could simply adhere to the position advanced by the US lawyers during the litigation. It did not.

14 In general, there is nothing like a clear statement by anyone (legislator or adjudicator) as to the legal value of WTO secondary law. It seems safe to argue, however, that they constitute an interpretative element of the primary law since they have been used as such by Panels so far.

15 By the same token, previous reports, such as the Panel Report on EC – Pipe Fittings, have used a recommendation by the ADP Committee as a source of law to reach its conclusion that it is desirable that the period for dumping investigation and for injury determination substantially overlap (§7.321).

16 Panel Report, Mexico – Anti-Dumping Measures on Rice, §7.82.

17 Panel Report, Mexico – Anti-Dumping Measures on Rice, §7.79. The Panel added, however, that its ruling should not be read as to imply that there could never be any convincing and valid reasons for examining only parts of years. Panel Report, Mexico – Anti-Dumping Measures on Rice, §7.82.

18 Panel Report, Mexico – Anti-Dumping Measures on Rice, §7.86.

19 Panel Report, Mexico – Anti-Dumping Measures on Rice, §7.57.

20 Panel Report, Mexico – Anti-Dumping Measures on Rice, §7.58.

21 Panel Report, Mexico – Anti-Dumping Measures on Rice, §7.63.

22 Panel Report, Mexico – Anti-Dumping Measures on Rice, §7.64.

23 Before the AB, Mexico complained that the Panel wrongly relied on the ADP recommendation. The AB rejected Mexico's argument, holding that the Panel relied on the recommendation only to confirm its own interpretation. This view seems to suggest that reliance on such instruments as a means to confirm already adopted interpretations is legally correct in the AB's view. If yes, this would ipso facto lead to classification of such elements as supplementary means of interpretation (in the sense of Art. 32 of the Vienna Convention on the Law of Treaties).

24 AB Report, Mexico – Anti-Dumping Measures on Rice, §§163–172.

25 AB Report, Mexico – Anti-Dumping Measures on Rice, §165.

26 Art. 2.1 AD, emphasis added.

27 Varian (Reference Varian1980: 651).

28 Stigler (Reference Stigler1961). See Baye, Morgan, and Scholten (Reference Baye, John, Patrick and Terry2006).

29 Art. 5.2, emphasis added.

30 Art. 6.10, emphasis added.

31 Art. 9.5 AD.

32 Art. 6.1 AD.

33 Art. 6.11 AD.

34 Art. 6.8 AD.

35 Although the legal status of Recommendations by WTO organs (and, indeed, of all secondary law) remains in limbo, dispute settlement practice evidences increasing reliance on such instruments: recommendations dealing with the length of the period of investigation offer an excellent illustration to this effect.


Baye, Michael, John, Morgan, and Patrick, Scholten (2006), ‘Information, Search, and Price Dispersion’, in Terry, Hendershott (ed.) Handbook of Economics and Information Systems, Amsterdam: Elsevier Science.Google Scholar
Stigler, George J. (1961), ‘The Economics of Information’, The Journal of Political Economy, 69 (3): 213225.CrossRefGoogle Scholar
Varian, Hal R. (1980), ‘A Model of Sales’, The American Economic Review, 70 (4): 651659.Google Scholar
You have Access
Cited by

Save article to Kindle

To save this article to your Kindle, first ensure is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about saving to your Kindle.

Note you can select to save to either the or variations. ‘’ emails are free but can only be saved to your device when it is connected to wi-fi. ‘’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

Find out more about the Kindle Personal Document Service.

Mexico – Antidumping Measures on Rice
Available formats

Save article to Dropbox

To save this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you used this feature, you will be asked to authorise Cambridge Core to connect with your Dropbox account. Find out more about saving content to Dropbox.

Mexico – Antidumping Measures on Rice
Available formats

Save article to Google Drive

To save this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you used this feature, you will be asked to authorise Cambridge Core to connect with your Google Drive account. Find out more about saving content to Google Drive.

Mexico – Antidumping Measures on Rice
Available formats

Reply to: Submit a response

Please enter your response.

Your details

Please enter a valid email address.

Conflicting interests

Do you have any conflicting interests? *