Appeals after the Appellate Body

Abstract

The Appellate Body (AB) of the World Trade Organization (WTO) has not heard an appeal since 2019. This article explores how adjudicators and member states have navigated WTO dispute settlement in this post-AB world. It begins by providing an overview of dispute settlement practice from 2020 to 2022, including by cataloguing appeals into the void, appeals to arbitration, and appeals forewent. It explains the incentives created by the lack of a functioning appeals mechanism and provides background on the alternative appeals procedure agreed to among a subset of WTO members: the Multi-Party Interim Arbitration Arrangement (MPIA). Moreover, it closely examines five WTO disputes: Colombia–Frozen Fries, Turkey–Pharmaceutical Products, EU–Steel Safeguards, Thailand–Cigarettes, and Costa Rica–Avocados. Through these five disputes, the article examines the circumstances in which members have agreed to binding appeals arbitration even absent formally committing to the MPIA, the circumstances in which members have appealed to arbitration or foregone such appeals, and whether facilitated negotiations present a workable alternative to an effective appeals mechanism. Finally, this article closely analyzes the reasoning of two appeals arbitration awards issued to date – Colombia–Frozen Fries and Turkey–Pharmaceutical Products – with a special focus on how those awards depart from AB precedent and what those departures can tell us about the current crisis.

1. Introduction

The Appellate Body (AB) of the World Trade Organization (WTO) has not heard an appeal since 2019. That is when the seven-member Body, which requires panels of three to hear appeals, dwindled to two, then to one, and then to none, as the US repeatedly blocked new member appointments to fill the Body's expiring ones. The aim of this article is to explore – through the close examination of five WTO disputes – how adjudicators and member states have navigated WTO dispute resolution in the absence of a functioning Appellate Body.

The WTO judiciary crisis that dismembered the AB has left the first instance adjudication of disputes formally intact. Member states of the WTO can still request the establishment of panels, whose ad hoc composition does not depend upon unanimous consensus of the entire WTO membership (as the AB does). Those panels, once established, can hear and decide disputes in the same way that they have since the WTO's founding in 1995 (and the founding of the General Agreement on Tariffs and Trade (GATT) in 1947 before that). Moreover, as in the past, disappointed members can still appeal adverse panel reports to the AB. However, today, any panel reports appealed to the AB simply join a long, stationary, and growing queue. That is problematic, because the rules of the WTO only allow for prospective remedies, and those remedies are only authorized after a dispute has run through the full appeals process. Appeals to the AB thus now provide members a legal way to avoid adverse panel reports and the consequences that flow from them. It allows members to appeal adverse reports ‘into the void’.

While the US has created this void, other members have sought to fill it. In April 2020, nineteen members of the WTO announced the creation of the Multi-Party Interim Arbitration Arrangement (MPIA or Arrangement).¹ The purpose of the MPIA is ‘to preserve … two levels of adjudication through an independent and impartial appellate review of panel reports’.² Accordingly, the MPIA commits its members to ‘resort to arbitration … as an interim appeal arbitration procedure … as long as the Appellate Body is not able to hear appeals of panel reports in disputes among them due to an insufficient number of Appellate Body members’.³ As of the end of 2023, the MPIA counts twenty-six members – including the EU and China, but not the US.⁴

Dispute resolution in the WTO is therefore now segmented. Any member can request the establishment of a first instance panel. However, among members of the MPIA, appeals of panel reports are subject to mandatory arbitration, while among non-MPIA members (or among one MPIA member and one non-MPIA member), appeals are subject to the void. The number of disputes that have been adjudicated in this new reality is relatively small, and certainly too small to draw any definite conclusions. Nevertheless, the hope of this article is that – even from this small sample – we can make some tentative observations about how states and adjudicators are navigating the Appellate Body crisis.

The remainder of this article is organized as follows: Section 2 presents a high-level overview of the panel reports issued from 2020 through the end of 2022 that were appealed into the void. Section 3 explores in detail two disputes (Colombia–Frozen Fries ⁵ and Turkey–Pharmaceutical Products)⁶ whose appeals were subject to arbitration, with a focus on how the resulting arbitration awards departed from prior AB reports. Section 4 explores two additional disputes EU–Steel Safeguards ⁷ and Cost Rica–Avocados ⁸, and the circumstances in which the parties forewent appeals despite an agreement to arbitrate. Section 5 discusses an additional dispute (Thailand–Cigarettes),⁹ in which the parties engaged a facilitator to help them achieve a settlement. The Conclusion then presents some tentative observations about WTO dispute settlement since the AB crisis.

2. Appeals into the Void

From 2020 through the end of 2022, panels have issued twenty-five reports.¹⁰ Of those reports, eighteen, or 72%, have been appealed into the void, to the now defunct AB. Another two have been appealed to arbitration. The overall appeal rate (20/25 or 80%) is consistent with recent rates of appeals before the demise of the Appellate Body.¹¹ However, appeals now have very different implications, given that the vast majority are filed with a body without the capacity to hear them.

Indeed, the numbers show that WTO members not bound by the MPIA have shown little self-restraint in the face of the void. Of the twenty-five panel reports issued, twenty-two were issued in disputes not subject to the mandatory arbitration agreement of the MPIA. Parties filed appeals into the void in eighteen of those twenty-two (~82%) disputes. In other words, where appealing into the void was an option, one of the parties did so eight out of ten times.

The lack of an AB has asymmetrical effects. On the one hand, where a panel agrees with the complainant and finds that the respondent is acting inconsistently with their WTO obligations, the respondent may vitiate that finding by appealing it into the void. On the other hand, where the panel disagrees with the complainant and finds that the respondent is acting consistently with its WTO obligations, neither party benefits from an appeal. The respondent has no reason to challenge a report that vindicates it, and the complainant gains nothing from appealing a report to a body that cannot reverse it. The Appellate Body crisis, thus, creates a situation whereby ‘heads’ respondents win, ‘tails’ complainants lose.

The appeals filed to date reflect this dynamic. Seventeen of eighteen appeals into the void were filed by the respondents. Only one panel report – in US–Photovoltaic Safeguard – was appealed by the complainant (in this case by China).¹² By contrast, where the parties were bound to appeal via arbitration (whether through the MPIA or a bilateral arbitration agreement), respondents appealed their losses in only two of five disputes, supporting (anecdotally at least) that appealing into the void is preferable to mandatory arbitration from a respondent's perspective.

Of course, today's respondent may be tomorrow's complainant. Ex ante, therefore, members have a reason to ensure that future disputes are resolved in a way that treats both respondents and complainants fairly. But the above dynamic and numbers underscore that once a dispute crystallizes, members' interests in effective third-party dispute resolution will diverge, and agreements to arbitrate appeals will become harder to achieve.

The MPIA – which demands that members pre-consent to arbitrate appeals in future disputes with other members – is responsive to this dynamic. However, that agreement still leaves MPIA members free to launch appeals into the void where the other party to the dispute is not an MPIA member. Accordingly, three of the appeals into the void have been made by MPIA members against non-MPIA members. In particular, the EU, Pakistan, and China have each appealed one report (against Russia, the United Arab Emirates, and the US, respectively). By contrast, fifteen of the eighteen appeals into the void were filed by non-MPIA members: the US has appealed eight reports into the void (three against China, and one each against Canada, Korea, Norway, Switzerland, and Turkey); India has appealed three reports (one against each Guatemala, Brazil, and Australia); and Morocco, Korea, Saudi Arabia, and Indonesia have each appealed one report (against Tunisia, Japan, Qatar, and the EU, respectively).

The evidence that the AB crisis impacts the number of disputes brought before the WTO is mixed. On the one hand, from 2020 to 2022, only twenty-two consultations have been requested, or on average about seven requests per year.¹³ While arguably a ‘steady stream of disputes’,¹⁴ it represents about a third of the WTO's caseload during the preceding twenty-five years, where there was an average of twenty-four disputes initiated per year.¹⁵ There may be several factors behind the increased reluctance to commence dispute resolution proceedings, but the threat of a complainant's victory at the panel stage being rendered futile by an appeal into the void is almost certainly one of them. On the other hand, fourteen of the twenty-two disputes launched since 2020 have involved at least one non-MPIA member, meaning that consultations were requested despite the (apparently high) risk that any eventual panel report would be appealed into the void.¹⁶

3. Appeals to Arbitration

From 2020 to 2022, two panel reports have been appealed to arbitration. However, only one such appeal (Colombia–Frozen Fries) was among members of MPIA (Colombia and the EU). The other (Turkey–Pharmaceutical Products) involved one MPIA member (EU) and one non-member (Turkey), who together entered into an ad hoc agreement to arbitrate the appeal in that dispute. This section summarizes these disputes with a focus on where their arbitration awards meaningfully departed from AB precedent.

3.1 The Arbitration Procedures

Both the appeals in Colombia–Frozen Fries and Turkey–Pharmaceutical Products were governed by a similar set of procedures defined by Annex 1 to the MPIA. Among other things, the Agreed Procedures require that ‘an appeal … be limited to issues of law covered by the panel report and legal interpretations developed by the panel’.¹⁷ They moreover require arbitrators ‘only address those issues that are necessary for the resolution of the dispute [and] only those issues that have been raised by the parties, without prejudice to their obligation to rule on jurisdictional issues’.¹⁸ The procedures of the MPIA further ‘request the arbitrators to issue the award within 90 days following the filing of the Notice of Appeal’, and give arbitrators the authority to ‘take appropriate organizational measures to streamline the proceedings’.¹⁹ These procedural rules are responsive to critiques levied against the AB by the US, including that it continuously exceeded the 90-day deadline imposed on it by Dispute Settlement Understanding (DSU) Article 17.5;²⁰ that the AB violated DSU Article 3.7 and WTO Agreement Article IX:2 by issuing advisory opinions on matters not necessary to resolve disputes;²¹ and that the AB violated DSU Article 17.6 by reviewing panels' factual findings.²²

In both Colombia–Frozen Fries and Turkey–Pharmaceutical Products, both arbitral panels issued their award within 90 days of the commencement of arbitration.²³ Moreover, in accordance with the procedural rules, the arbitrators limited their awards to ‘issues of law’ and did not revisit factual findings by the panel in any obvious way. Whether they limited their decisions to ‘those issues that are necessary for the resolution of the disputes’ is less clear, as discussed further below. As further envisioned by the MPIA, the arbitrators in both disputes received assistance in the form of legal advisors from the WTO Secretariat.²⁴ The legal advisors were selected on an ad hoc basis to ensure they had no previous affiliation with the disputes (either because they served as part of the Legal Affairs or Rules Divisions that assist panels, or in any other way).²⁵

The main procedural differences between the two appeals arbitrations involved the appointment of the arbitrators. In Colombia–Frozen Fries, the arbitrators were selected according to a process dictated by the MPIA.²⁶ In particular, the MPIA establishes that each dispute be resolved by three arbitrators selected from a ‘pool of 10 standing appeal arbitrators’.²⁷ That pool is established by way of consensus of the MPIA members, based on the nominations of participating members (each of whom may nominate one candidate).²⁸ The MPIA calls for arbitrators to be selected from the pool ‘on the basis of the same principles and methods that apply to form a division of the Appellate Body under Article 17.1 of the DSU and Rule 6(2) of the Working Procedures for Appellate Review, including the principle of rotation’.²⁹ By contrast, in Turkey–Pharmaceutical Products, Turkey and the EU agreed to a bespoke selection procedure that required that the arbitral tribunal represent a mix of former AB Members and MPIA appeal arbitrators,³⁰ perhaps to reflect the fact that Turkey (not an MPIA member) had played no role in selecting the MPIA pool of arbitrators.

3.2 MPIA Arbitration: Colombia–Frozen Fries

Colombia–Frozen Fries, at the time of this writing, was the first and only WTO appeal among MPIA members to be resolved through arbitration.

3.2.1 The Panel Report

The dispute arose from antidumping duties imposed by Colombia on certain frozen potato products originating from Belgium, Germany, and the Netherlands. In February 2020, the EU had requested the establishment of a panel, which issued its report to the parties in August 2022. The panel report found, among other things, that Colombia had acted inconsistently with Article 5 of the Anti-Dumping Agreement (ADA) in initiating its antidumping investigation.³¹ More particularly, the panel found that the Colombia's national authority improperly initiated its investigation based upon an industry application that used third-country sales prices (here, the price of the products in the UK) as the basis for calculating dumping margins.

Put simply, dumping margins are calculated by subtracting the export price of a product from its ‘normal value’. Where the export price is lower than the normal value, dumping is said to occur, and the difference in price amounts to the dumping margin. That margin is then used to calculate antidumping duties (i.e., duties levied on the imported products meant to counter-act the dumping). Members wishing to impose higher anti-dumping duties to protect domestic industry face an incentive to inflate the normal value to increase the dumping margin and anti-dumping duties. The ADA accordingly imposes some requirements for determining normal value.

Colombia argued that the ADA granted it a ‘free choice’ to use third-country sale prices as opposed to the products' domestic sales prices (i.e., the prices the products sold in Belgium, Germany, and the Netherlands) in its calculation of the products' normal value.³² According to the panel, however, the ADA allows the use of third-country prices only where ‘appropriate’. Moreover, Article 5 requires that a national authority ‘examine the “adequacy” of the evidence contained in the application to determine whether there was “sufficient evidence” to justify initiation’ of an investigation.³³ Here, because neither the application nor the national authority explained why the use of third-country sales prices was ‘appropriate’, the panel found that Colombia had acted inconsistently with its ADA obligations in its antidumping investigation.³⁴

3.2.2 The Arbitration Award

Colombia appealed the panel report pursuant to the MPIA Arrangement. In their award, the arbitrators addressed several issues raised on appeal by Colombia, including claims under Articles 3 and 6 of the ADA. This Section focuses specifically on the arbitrators' interpretation of ADA Article 17.6(ii), which (according to the arbitrators) was critical to the interpretation and application of Article 5,³⁵ and which departed from prior AB reports.

Article 17.6(ii) of the ADA provides:

In examining the matter referred to in paragraph 5:

…

(ii) the panel shall interpret the relevant provisions of the Agreement in accordance with customary rules of interpretation of public international law. Where the panel finds that a relevant provision of the Agreement admits of more than one permissible interpretation, the panel shall find the authorities' measure to be in conformity with the Agreement if it rests upon one of those permissible interpretations.

The panel had determined that Colombia's interpretation of Article 5 was not ‘permissible’ under Article 17.6(ii) by first conducting its own interpretation of Article 5, and then assessing whether that process allowed the interpretation reached by Colombia.³⁶ That approach was consistent with several previous decisions of the Appellate Body. For example, in EC–Bed Linens, the Appellate Body upheld a panel's interpretation that first interpreted the ADA according to the Vienna Convention, and then found that that the European Community's interpretation of the same term was not permissible.³⁷ In US–Continued Zeroing, the AB expressly explained that ‘Article 17.6(ii) contemplates [that] sequential analysis’:

The first step requires a panel to apply the customary rules of interpretation to the treaty to see what is yielded by a conscientious application of such rules including those codified in the Vienna Convention. Only after engaging this exercise will a panel be able to determine whether the second sentence of Article 17.6(ii) applies. The structure and logic of Article 17.6(ii) therefore do not permit a panel to determine first whether an interpretation is permissible under the second sentence and then to seek validation of that permissibility by recourse to the first sentence.³⁸

The arbitrators in Colombia–Frozen Fries departed from that approach. They reasoned that Article 17.6(ii) precluded the tribunal from ‘engag[ing] in [its] own, de novo interpretation’ of Article 5 ‘so as to arrive at what [the tribunal] consider[s] to be the “final” or “correct”’ interpretation.³⁹ Instead, Article 17.6(ii) requires asking ‘whether a treaty interpreter, using the method for treaty interpretation set out in the Vienna Convention … could have reached Colombia's interpretation … even though [the arbitrators], as de novo treaty interpreters, might have reached a different conclusion’.⁴⁰ In other words, the ‘starting point’ in the assessment of whether the member's interpretation (here Colombia's) is ‘permissible’ under the second sentence of Article 17.6(ii), is the member's interpretation – not the panel's.⁴¹

According to the arbitrators, several factors justify this approach. First, the arbitrators argued that the second sentence of 17.6(ii), read in the context of 17.6(i) ‘must be understood in a manner granting special deference to investigating authorities under the Anti-Dumping Agreement’.⁴² In particular, Article 17.6(i) ‘prevents a panel from conducting a de novo assessment of the facts on record; [rather] an authority's establishment and evaluation of facts must be allowed to stand so long as it is “proper” and “unbiased and objective”, and this is the case “even though the panel might have reached a different conclusion”’.⁴³ According to the arbitrators, Article 17.6(ii) should be read as calling for a parallel type of deferential review: not a de novo interpretation of the law on record, but rather an assessment of whether the member's own interpretation was ‘permissible’.

Second, the arbitrators emphasized ‘[t]reaty interpretation is not an exact science and applying the Vienna Convention's method does not magically and inevitably lead to a single result. In most cases, treaty interpretation involves weighing, balancing, and choice’.⁴⁴ Accordingly, ‘different treaty interpreters applying the same tools of the Vienna Convention may, in good faith and with solid arguments in support, reach different conclusions on the “correct” interpretation of a treaty provision’.⁴⁵ According to the arbitrators, the Anti-Dumping Agreement was ‘particularly’ susceptible to multiple good faith differing interpretations given that it was ‘drafted with the understanding that investigating authorities employ different methodologies and approaches’.⁴⁶

Ultimately, however, while the arbitrators' interpretive approach differed from the panel's, its conclusion did not. The arbitrators, like the panel, found that Colombia's interpretation of Article 5 was not ‘permissible’ under Article 17.6(ii). The tribunal reasoned as follows:

Article 5.2(iii) specifies three types of product prices for normal value: (i) domestic sales; (ii) third-country sales; and (iii) constructed normal value. The structure of the sentence, however, does not place these three sources of prices on equal footing. Rather than simply listing the three in sequence, third-country sales prices and constructed normal value, but not domestic sales prices, are qualified by the phrase ‘where appropriate’.⁴⁷

…

The fact that the phrase ‘where appropriate’ only applies to third-country sales prices and constructed normal value signifies that only these prices [and not domestic prices] require an authority to ‘examine’ their ‘appropriateness' to initiate an investigation under Article 5.3.⁴⁸

Accordingly, the arbitrators concluded that Colombia's interpretation of Article 5.3 – according to which it had a ‘free choice’ to use third country prices without explanation – was not ‘permissible’. And because the facts on record showed that the Colombian authority failed to ‘examine’ whether the use of third-country prices was ‘appropriate’, it acted inconsistently with Article 5.3.⁴⁹

3.2.3 Observations

The arbitrators' departure from past Appellate Body interpretations of Article 17.6 is remarkable for a number of reasons, including most obviously that it is strained from a textual perspective. Here again is the text of Article 17.6(ii):

the panel shall interpret the relevant provisions of the Agreement in accordance with customary rules of interpretation of public international law. Where the panel finds that a relevant provision of the Agreement admits of more than one permissible interpretation, the panel shall find the authorities’ measure to be in conformity with the Agreement if it rests upon one of those permissible interpretations.

As several Appellate Body reports (and the panel in this dispute) explained, the first sentence commands panels (‘shall’) to ‘interpret the ADA in accordance with customary rules of interpretation of public international law’. It is hard to read that sentence as anything less than a requirement that a panel ‘interpret the ADA’ itself (i.e., conduct a de novo review) according to the Vienna Convention. The second sentence has two clauses, the second of which was emphasized by the arbitrators and calls on the panel to ‘find the authorities' measure to be in conformity with the Agreement if it rests upon [a] permissible interpretation’. But the first clause of that sentence makes clear that the panel's determination of what is permissible is evaluated with respect to the panel's interpretation. That clause provides: ‘Where the panel finds that a relevant provision of the Agreement admits of more than one permissible interpretation … ’. Contrary to the arbitrators' award, the structure of Article 17.6(ii) quite clearly establishes the panel's interpretation as the ‘starting point’.

Several commentators have argued that this reading of Article 17.6(ii) ‘ignores’ or ‘read[s] out of the law the deference required to domestic authorities’.⁵⁰ The idea appears to be that Article 17.6(ii)'s call for the panel to interpret the ADA according to the Vienna Convention leaves that Article doing nothing at all, or at least nothing that is not already required by customary international law or the Dispute Settlement Understanding.⁵¹ But a persuasive answer to that critique can be found within the Colombia–Frozen Fries arbitration award itself, in which the arbitrators note that ‘different treaty interpreters applying the same tools of the Vienna Convention may … reach different conclusions on the “correct” interpretation of a treaty provision’.⁵² One interpretation of Article 17.6(ii) that respects its plain text and gives it effect is to read it as an acknowledgement that the ADA is susceptible to multiple permissible interpretations, and as a call to panels to be open to finding them when conducting their de novo interpretation.⁵³

There is another (non-mutually exclusive) way to interpret Article 17.6(ii) that gives it effect while respecting its text: as an implicit limit on the appellate review of ADA disputes. Indeed, if one takes seriously the text of Article 17.6(ii), the role of the Appellate Body is not to conduct its own de novo interpretation of the ADA, nor is it even to determine whether the member's interpretation is ‘permissible’. Rather, the role of the Appellate Body (or an arbitral stand-in for that body) is simply to determine whether the panel carried out those tasks. Where the AB (or a stand-in arbitration panel) finds that the panel did so, as was arguably the case here, that is the end of the story. Where the AB finds that the panel did not, only then can it correct the panel report by standing in place of the panel. According to this interpretation, Article 17.6(ii) establishes deference. But in the context of an appeal, that deference is to the panel's interpretation of the ADA.

The arbitrators here, of course, did not take that approach, choosing instead to place themselves in the panel's position, as the AB had done many times before. This time, however, they sat with a different interpretation of what Article 17.6(ii) required of them and of the panel whose report they were reviewing. As discussed above, the arbitrators emphasized that Article 17.6(ii) precluded them from ‘engag[ing] in [their] own, de novo interpretation’ of Article 5 ‘so as to arrive at what [the tribunal] consider[s] to be the “final” or “correct”’ interpretation.⁵⁴ This was ostensibly the reason why they were reviewing the panel's findings: it had impermissibly conducted its own de novo interpretation of Article 5. But in their attempt to correct the panel's error, the arbitrators ended up making the exact same ‘mistake’. Yes, they started with Colombia's interpretation that Article 5 affords them a ‘free choice’ to reference domestic and third country party prices without needing to explain the choice. And they ended by concluding that they could not ‘accept Colombia's interpretation as a “permissible” one’.⁵⁵ But they arrived at that conclusion only after they themselves affirmatively interpreted Article 5 as requiring that an authority ‘examine’ the ‘appropriateness’ of third party prices before initiating an investigation under Article 5.3.⁵⁶ That affirmative interpretation was the reference point for assessing Colombia's interpretation and concluding that it was not ‘permissible’. It was, for all intents and purposes, identical to the interpretation issued by the panel.

And that raises the question of why the arbitrators bothered departing from AB precedent on Article 17.6 at all. The question is all the more interesting because they could have (and arguably should have) avoided offering their own interpretation of Article 17.6(ii) altogether. The Agreed Procedures for arbitration under the MPIA provide that ‘[t]he arbitrators shall only address those issues that are necessary for the resolution of the dispute [and] only those issues that have been raised by the parties … ’.⁵⁷ Here, the arbitrators' interpretation of Article 17.6(ii) was arguably not ‘necessary for the resolution of the dispute’, since it had no impact on the outcome. Moreover, Colombia had not ‘raised’ the panel's interpretation of 17.6(ii) as an ‘issue’ in its appeal.⁵⁸ The arbitrators justified their focus on Article 17.6(ii) on the grounds that ‘whether the Panel's specific findings under Articles 5.2(iii) and 5.3 constitute[d] a legal error must be guided by Article 17.6’. But even if that was the case, the arbitrators could have resolved the dispute without issuing or defending a definitive interpretation of Article 17.6(ii). They could have instead acknowledged two competing interpretations of the Article and concluded, without prejudice as to which interpretation was correct, that Colombia was in violation of Article 5 under either. Put simply, they could have, and arguably should have, avoided opining on the interpretation of Article 17.6(ii).

Why didn't they? The most obvious explanation is that they simply became caught up trying to make sense of a notoriously enigmatic provision. But it is also hard to miss that Article 17.6(ii) was a flashpoint that preceded (and arguably led to) the present AB crisis. More specifically, the Appellate Body's interpretation of Article 17.6(ii) was at issue in several antidumping disputes involving the United States' practice of ‘zeroing’. As discussed above, dumping refers to the practice of exporting products at prices below their normal value (which is assessed with respect to domestic prices or where appropriate third country prices). Zeroing is the practice of ignoring negative dumping margins – meaning those imports whose prices exceed the normal value of a product. That practice has the effect of increasing dumping margins as compared to calculations that incorporate those negative margins, because it effectively raises negative margins to ‘zero’. The AB has ‘consistently found that zeroing is illegal’,⁵⁹ because it has repeatedly found that the US's interpretation of the ADA, on which its application of zeroing rests, is not a ‘permissible’ one under Article 17.6(ii).⁶⁰

The AB's rulings on zeroing are believed to have been critical to the US's decision to cripple the AB. The United States Trade Representative's (USTR) 2020 Report on the Appellate Body ‘excoriat[ed]’ the AB's decisions on zeroing,⁶¹ singling out the AB's ‘failure to give meaning to Article 17.6(ii) of the Antidumping Agreement’ as ‘exacerbating’ the ‘erroneous’, and ‘flawed interpretations’⁶² that led to the ‘invention of [it]s prohibition’.⁶³ If the AB gave that clause meaning, according to the US, it would accept that the ADA could be read in multiple ways, including one that allowed zeroing. In recognition of the importance of this issue to the AB's revival, Ambassador David Walker's draft decision on the functioning of the Appellate Body expressly singled out Article 17.6(ii) under the category of ‘overreach’, calling on panels and the AB to ‘interpret provisions of the [ADA] in accordance with Article 17.6(ii) of that Agreement’.⁶⁴

Whether purposeful or not, the arbitrators’ interpretation of Article 17.6(ii) in Colombia–Frozen Fries is responsive to these critiques and recommendations. It sends a signal to other adjudicators that a course correction on Article 17.6(ii) is appropriate. And it sends a signal to the US that at least some adjudicators are willing to make that correction. Of course, nothing in the arbitrators' interpretation of Article 17.6(ii) guarantees the US's desired substantive outcomes with respect to zeroing. As Colombia–Frozen Fries demonstrates, a liberal interpretation of Article 17.6(ii), which ostensibly grants broad deference to the interpretation of national authorities, in no way guarantees that those authorities' interpretations will in fact be found ‘permissible’. Ultimately, that question can only be answered, one way or another, by a panel or an appeals body, by reference to the text of the ADA itself.

3.3 Quasi MPIA: Turkey—Pharmaceutical Products

The second panel report to be appealed pursuant to Article 25 arbitration was Turkey—Pharmaceutical Products. In contrast to Colombia – Frozen Fries, this dispute involved only one member of the MPIA (the EU), and thus the mandatory arbitration provision of the MPIA Arrangement did not apply. Recourse to arbitration in Turkey–Pharmaceutical Products was instead established by an ad hoc bilateral agreement between the parties. That agreement was notified to the Dispute Settlement Body (DSB) after the parties had received the panel report, and after they would have seen that Turkey was found to have acted inconsistently with GATT Article III.⁶⁵ Interestingly, on the same day, the parties also notified the DSB of their agreement to arbitrate any appeal in a different dispute (EU–Steel Safeguards), in which the EU was found to have acted inconsistently with its obligations.⁶⁶ As discussed below, no appeal in that latter dispute was filed, however.

3.3.1 The Panel Report

The dispute in Turkey–Pharmaceutical Products arose from a Turkish measure that required ‘foreign producers to commit to localize in Turkey their production of certain pharmaceutical products’, in order to benefit from Turkey's Social Security Institution's (SSI) reimbursement program. The program reimbursed retail pharmacies for part of the price of certain pharmaceutical products distributed to outpatients.⁶⁷ Where ‘commitments are not given, accepted, or fulfilled’, by the foreign producers to localize their production, their ‘products are no longer reimbursed by the SSI’.⁶⁸ The stated aim of the Turkish government was to ‘to meet 60% of domestic pharmaceutical demand through domestic production’, to protect against supply chain disruptions, and to ‘ensure uninterrupted access to safe, effective, and affordable pharmaceutical products for all patients in Turkey’.⁶⁹

The panel agreed with the EU that this requirement constituted a prima facie breach of national treatment obligations under Article III of GATT. The panel further found – contrary to what Turkey argued – that the localization requirement was not covered by the government procurement derogation in Article III:8(a) of the GATT.⁷⁰ It went on to examine whether that violation was excused by Article XX(b) or XX(d) of the GATT and found that it was not.

3.3.2 The Arbitration Award

On appeal, Turkey challenged the panel's interpretation and application of Article III:8(a) and of Article XX(b) and (d).⁷¹ This Section focuses on the arbitrators’ interpretation of Article III:8, which was critical to the arbitration award, and which departed from prior case law.

Article III:8(a) of the GATT establishes a derogation from the national treatment obligation for measures relating to government procurement. It provides:

The provisions of this Article shall not apply to laws, regulations or requirements governing the procurement by governmental agencies of products purchased for governmental purposes and not with a view to commercial resale or with a view to use in the production of goods for commercial sale.

The panel had found that in order to fall within the derogation under III:8(a), the measure ‘must involve a “purchase” of products by a “governmental agency”’.⁷² And here, because the pharmaceutical products were not purchased by governmental agencies (the government only reimbursed the cost)⁷³ the localization requirement did not fall within the derogation.⁷⁴

The panel's interpretation of Article III:8(a) was consistent with (limited) AB precedent on the issue. Most notably, in India–Solar Cells the Appellate Body held:

The measures within the scope of Article III:8(a) are ‘laws, regulations or requirements governing … procurement’, and the entity purchasing products needs to be a ‘governmental agency’. Furthermore, the scope of Article III:8(a) is limited to ‘products purchased for governmental purposes’, and ‘not with a view to commercial resale or with a view to use in the production of goods for commercial sale’.⁷⁵

The panel's interpretation was also consistent with the AB's reports in Canada – Renewable Energy Generation Sector, and Canada–Feed-in-Tariff Program. In those reports, the AB distinguished between the meanings of ‘procurement’ and ‘purchased’,⁷⁶ but concluded ‘[t]he provision exempts from the national treatment obligation certain measures containing rules for the process by which government purchases products’.⁷⁷

The arbitrators departed from these interpretations and overturned the panel's conclusion that Article III:8(a) required a ‘purchase by a governmental agency’. The arbitrators' interpretation rested on two conclusions. First, the different terms ‘procurement’ and ‘purchase’ in Article III:8(a) must mean different things. According to the arbitrators, while procurement may involve a purchase, it need not.⁷⁸ A governmental agency can procure products by, for example, renting or leasing.⁷⁹ Procurement, however, does require ‘a certain level of control over the products purchased for governmental purposes’, and ‘acquiring or obtaining products cannot be equated with merely financing or regulating the acquisition of products’.⁸⁰ Second, the arbitrators found that because the term ‘by government agencies’ follows the term ‘procurement’, and not the term ‘purchased’, ‘by government agencies’ qualifies the former and not the latter. As such, in order to fall within the derogation, the government agency need only ‘procure’ the products, not ‘purchase’ them. The arbitrators concluded that the ‘relevant purchase transaction may be entered into by a non-governmental entity so long as the products are procured by a governmental agency and procurement is of products purchased for governmental purposes’.⁸¹

The arbitrators' divergent interpretation of Article III:8 had no effect on the ultimate outcome, however. The arbitrators, like the panel before it, found that the pharmaceutical products were not in any event ‘procured’ by the SSI (the Turkish governmental agency). Adopting the facts as determined by the panel, the arbitrators noted that the SSI never took physical possession or ownership of the products;⁸² never disposed of or controlled the products; never benefited from their use or managed their stocks,⁸³ and never acquired the rights to do those things. Rather, all of those rights remained with wholesalers, the pharmacies, and the patients.⁸⁴ In light of that, the arbitrators concluded that ‘[t]he fact that the SSI decides which pharmaceutical products are included in the Annex 4/A list and sets their price, enters into individual contracts with retail pharmacies, and pays the invoices that are periodically sent by the retail pharmacies does not show … that there is procurement by the SSI’.⁸⁵ Moreover, the arbitrators upheld the panel's finding that the reimbursement program did not fall within the public policy exceptions of GATT Article XX.

3.3.3 Observations

The arbitrators' decision in this dispute is remarkable for some of the same reasons as the award in Colombia–Frozen Fries.

First, as in Colombia–Frozen Fries, the arbitrators' interpretation here is somewhat strained. They conclude that because ‘by governmental agencies’ follows the term ‘procurement’ and not ‘products purchased’, the government is only obligated to have engaged in the procurement, not the purchase, of products for the derogation to apply. That interpretation envisions circumstances in which one entity (a governmental agency) engages in the ‘procurement’ of products and another (a non-governmental agency) engages in the ‘purchase’ of those products. However, that begs the question: if the government is not required to purchase the products for its procurement to fall within the derogation, why would there be any requirement that a product be ‘purchased’? Put another way, if the government can procure products by way of a lease and still fall within the derogation, why would the derogation depend on the non-governmental entity purchasing them (as opposed to leasing or manufacturing them)? If the arbitrators' mechanical interpretation does not lead to ‘manifestly … unreasonable’ results,⁸⁶ it at least raises questions of coherence.⁸⁷

Second, as in Colombia–Frozen Fries, it is unclear whether the arbitrators' interpretation of Article III:8 was necessary for resolving the dispute before it. As in Colombia–Frozen Fries, the arbitrators here were bound by a requirement to ‘only address those issues that are necessary for the resolution of the dispute’ and ‘only those issues that have been raised by the parties’.⁸⁸ And here, as in Colombia–Frozen Fries, the arbitrators' divergent interpretation of Article III:8 had no impact on the ultimate ‘resolution of the dispute’. Despite finding that Article III:8 only required the government agency to have ‘procured’ but not ‘purchased’ the products, it concluded that, in this case, Turkey had in any event not procured the products even under its wider definition of procurement. In other words, the arbitrators here could have resolved the dispute by noting, but not deciding on, the divergent interpretations of Article III:8, and concluding that Turkey would be afoul under either interpretation.⁸⁹

Third, like in Colombia–Frozen Fries, the arbitrators here modified the legal interpretation of the panel in a way that loosens, if only slightly, the discipline of the WTO Agreements on its members. Under the arbitrators' reading, in contrast to the panel's, Article III:8's derogation from national treatment discipline expands to include ‘the procurement by government agencies of products’ regardless of whether they are purchased by a governmental agency or not. That interpretation will have no impact on the Agreement on Government Procurement (GPA), a plurilateral agreement which has twenty-one members, including the US and the EU, and which imposes additional requirements (including national treatment) among those members with respect to government procurement. In fact, the GPA already applies to procurement ‘by any contractual means, including: purchase; lease; and rental or hire purchase, with or without an option to buy’.⁹⁰ However, the arbitrators' interpretation, if followed by future panels, would loosen discipline on government procurement vis-à-vis non-members of the GPA, including China. As the United States forays deeper into industrial policy and indulges more in ‘Buy America’ provisions that favor domestic producers in government procurement contracts, that distinction may become more significant.⁹¹

4. Foregone Appeals

Five panel reports issued from 2020 to 2022 were not appealed at all – either into the void or to arbitration.⁹² Interestingly, the parties forewent appeals in three of these disputes notwithstanding agreements to arbitrate such appeals pursuant to DSU Article 25.⁹³ In this section, I explore two of those cases.

4.1 EU–Steel Safeguards

The dispute in EU–Steel Safeguards arose from safeguard measures imposed by the EU in the form of a combination of duty-free tariff rate quotas and 25% out-of-quota safeguard duties.⁹⁴ The EU applied the safeguards on the grounds that unforeseen developments had resulted in an increase in imports of steel into the EU, which was threatening serious injury to EU steel producers. The unforeseen developments included: (i) increased overcapacity in global steel production, (ii) an increase in the use of trade restrictive and trade defense measures on steel globally, and (iii) the US Section 232 measures on steel, which had imposed a 25% tariff on steel imports into the US.⁹⁵ Turkey challenged the EU's safeguard measures, and the panel found that the measures were in fact inconsistent with a number of obligations under the Agreement on Safeguards and the GATT.⁹⁶

Among other claims, the panel accepted Turkey's claim that the EU had not established that the increase in imports had taken place ‘as a result of’ the identified unforeseen developments. The panel reasoned ‘that the phrase “as a result of” does not establish a causation requirement’,⁹⁷ between the unforeseen development and increase in imports, but it does require showing ‘a logical connection between the [two]’.⁹⁸ Here the EU provided evidence of production overcapacity, but its analysis was limited to ‘asserting “it is clear” that overcapacity leads producers to seek other export opportunities, observing that import prices were typically lower than prices of EU producers, and concluding that this has resulted in increased imports into the European Union’.⁹⁹ That, according to the panel, was insufficient to establish the requisite connection as required under Article XIX:1(a),¹⁰⁰ which ‘require[s] a much more detailed analysis’ than ‘bringing two sets of facts together’.¹⁰¹

Additionally, the panel found that the EU's finding of a threat of serious injury was not ‘based on facts’ as required by Article 4.1(b) of the Agreement on Safeguards. The heart of the panel's finding rested on the observed improvement of the EU steel industry's performance¹⁰² toward the end of the period of investigation.¹⁰³ Here, the EU had argued that the industry's improvement could be explained by the imposition of anti-dumping duties and countervailing measures, and that the industry was therefore still vulnerable to serious injury. However, the panel found that explanation to be insufficient, in part because some improvement occurred in the face of increased imports that were not subject to either antidumping duties or countervailing measures.¹⁰⁴

Because Turkey is not a member of the MPIA, the EU had the option to appeal the adverse panel report into the void. However, as discussed above, at the time that the EU and Turkey entered into an agreement to arbitrate the appeal in Turkey–Pharmaceutical Products, they also agreed to arbitrate any appeals in EU–Steel Safeguards. Both arbitration agreements were made after the panel reports in each dispute had been circulated to the parties, i.e. after the parties had been notified that they had each lost one of the two disputes.¹⁰⁵ The former arbitration agreement paved the way for Turkey's appeal in Turkey–Pharmaceutical Products discussed above. By contrast, despite Article 25 arbitration being available to it, the EU chose not to appeal its loss in EU–Steel Safeguards. Instead, the panel report was adopted by the DSB on 31 May 2022.

Why the EU did not appeal its loss in EU–Steel Safeguards is anyone's guess. Perhaps the EU determined an appeal was hopeless or perhaps the EU decided the panel report was something it could live with. Indeed, several months after the panel report was adopted, the EU issued a revised definitive safeguard measure on steel.¹⁰⁶ But that begs the question of why the EU would go to the trouble of agreeing to arbitrate an appeal in this dispute in the first place, particularly since it did so after it would have already seen a draft of the panel report?¹⁰⁷ One possible answer is that the EU's agreement to arbitrate an appeal in EU–Steel Safeguards was made in exchange for Turkey's agreement to arbitrate its appeal in Turkey–Pharmaceutical Products. In each dispute, absent the agreement to arbitrate, each of the members would have had the option of appealing their respective losses into the void, thus vitiating the victory of the other member. In this context, the EU's agreement to arbitrate any appeal in EU–Steel Safeguards (and not appeal it into the void) could be understood as a concession made in exchange for Turkey's agreement to do the same in Turkey–Pharmaceutical Products.

A similar logic of reciprocity underpins the MPIA. MPIA members' pre-consent to arbitrate appeals (and thus forego appeals into the void) can be understood as a concession that protects the future victories of other MPIA members. In return, each member receives the same concession from other members. As discussed above, pre-consent is important because once a dispute crystallizes, the respondent will have an incentive to preserve its option to appeal into the void. By contrast, members' incentives to preserve the integrity of dispute resolution are more closely aligned before disputes crystallize, as each has an interest to preserve its capacity to bring effective claims as a complainant.

The agreement to arbitrate appeals in Turkey–Pharmaceutical Products and EU–Steel Safeguards shows, however, that reciprocity can be reproduced ex post in the context of parallel losses. In this context, each member can promise to refrain from appealing its own loss into the void in order to preserve its panel victory in the other dispute. While the parallel losses appear to be coincidental in the EU–Turkey disputes, complainants could also reproduce them on demand by taking inconsistent measures that invite counterclaims, which can then be used as leverage to protect the integrity of the original dispute. Whether (non-MPIA) complainants attempt to engineer such opportunities – and whether they can do so without spiraling into a trade war – is something to watch for.

4.2 Costa Rica –Avocados

In Costa Rica–Avocados, there was no need for Mexico, the complainant, to manufacture its own loss to protect its victory against Costa Rica, as both states were members of the MPIA. Notably, both states were part of the original nineteen members to announce the MPIA in 2020. That announcement was made after Mexico had requested the establishment of a panel in this dispute in 2018, but before the panel would circulate its final report in April 2022.¹⁰⁸ In other words, both states agreed to join the MPIA knowing that any appeals of the present dispute would be subject to arbitration. In the end, Mexico prevailed at the panel stage, and Costa Rica forewent an appeal, agreeing instead to implement the panel's recommendation within a reasonable period of time.

At the center of the dispute were restrictions on the importation of avocados from Mexico that were ostensibly designed to protect against avocado sunblotch viroid. Avocado sunblotch viroid (ASBVd) is a disease that affects avocado trees and fruit, including by decreasing the yield of avocado trees.¹⁰⁹ In order to protect its own avocado production, Costa Rica required that avocado fruit imports from Mexico be accompanied by an official certificate issued by Mexico indicating that the fruit was free of ASBVd; an official certificate issued by Mexico indicating the fruit came from a place of production that was free of ASBVd; or otherwise adhere to a bilaterally agreed program related to ASBVd.¹¹⁰ Costa Rica justified the measures on the grounds that the importation of Mexican avocados posed a risk of contaminating Costa Rican avocado trees with ASBVd and thus lowering Costa Rican avocado production.

The panel found that Costa Rica's measures were inconsistent with Articles 2 and 5 of the Agreement on the Application of Sanitary and Phytosanitary Measures (SPS Agreement) for several reasons. First, Costa Rica's determination that its territory was free of ASBVd (and thus could be protected by import restrictions) suffered from various flaws, including deficiencies in its sampling and surveillance of Costa Rican territory,¹¹¹ and that, as such, one of the bases of its risk assessment lacked reliability.¹¹² Second, the panel found that Costa Rica produced insufficient scientific evidence to show that imported avocado fruit posed a risk to Costa Rican avocado trees. Third, the panel found that Costa Rica failed to take into account the costs associated with the risk of contamination and the relative cost-effectiveness of alternatives.¹¹³ In conclusion, the panel found that Costa Rica's measures were not based on a risk assessment as required by SPS Article 5, nor were they based on scientific principles or scientific evidence as required by SPS Article 2.¹¹⁴

There is nothing terribly out of the ordinary about Cost Rica's decision not to appeal the panel's decision. During the first 25 years of the DSU, approximately 24% of panel reports went unappealed.¹¹⁵ Perhaps here, Costa Rica viewed the appeal as legally hopeless or in any event not beneficial from a pure cost–benefit perspective. Perhaps Costa Rica demurred for political reasons. Regardless of why Costa Rica declined to appeal, what is clear is that it would have faced a different calculus if it were not bound by the MPIA and appealing into the void had been an option.

5. Appointment of a Facilitator: Thailand–Cigarettes

In one additional dispute, ongoing since 2008, the parties employed another innovation to work-around the defunct Appellate Body. In 2020, Thailand and the Philippines agreed to select a ‘facilitator’ to help them resolve Thailand–Cigarettes, which had resulted in two adverse compliance reports that Thailand had appealed into the void.

5.1 Background

The dispute in Thailand–Cigarettes dates back to 2008, when the Philippines requested the formation of a panel to challenge the application of certain fiscal and taxation measures imposed by Thailand on imported cigarettes. The Philippines alleged, and the panel agreed, that various measures and their application violated both the GATT and the Customs Valuation Agreement (CVA). Among other things, the panel found that Thailand improperly rejected the importers' declared transaction value as the basis for calculating its customs duties in violation of Articles 1 and 16 of the CVA; calculated value added taxes (VAT) on imported cigarettes in a way that violated Articles III:2 and III:4 of the GATT; and failed to maintain an independent tribunal to review customs matters in violation of GATT Article X:3(b).¹¹⁶ Thailand appealed the panel report to the Appellate Body, which upheld its main conclusions in June 2011.¹¹⁷

In January 2013, Thailand notified the DSB that it had completed the implementation of the AB's recommendations. The Philippines, however, disagreed and after consultations requested the formation of a compliance panel in June 2016. In a report issued on in November 2018, the compliance panel found that Thailand was not in compliance with the AB report.¹¹⁸ In March 2018 – before the issuance of that (first) compliance report – the Philippines requested the establishment of a second compliance panel related to measures Thailand took after the first set of compliance consultations had taken place. That second compliance panel, which issued its report on in July 2019, also found that Thailand was out of compliance.¹¹⁹

In January 2019 and July 2019, respectively, Thailand notified the DSB of its appeals of the first and second compliance reports to the Appellate Body. In February 2020, the Philippines requested authorization from the DSB to suspend concessions pursuant to DSU Article 22.2. Thailand opposed the request on the grounds that the compliance reports were subject to appeal before the Appellate Body, which at this point lacked a quorum to hear disputes, and that the sequencing agreement entered into between it and the Philippines precluded suspending concessions until the appeals were decided.

5.2 Appointment of a Facilitator

In December 2020, after several months of back and forth, the two members issued a joint communication stating their desire to ‘further deepen [the] process of consultations, through the additional assistance of a Facilitator nominated by [then chairman of the DSB] Ambassador Castillo’, to achieve a comprehensive settlement between them.¹²⁰ The parties laid out the parameters of the facilitator's role and process, as follows:

The Facilitator, building on the clarification process initiated by the current DSB Chair, will seek to identify and make recommendations to the parties on ways and means of resolving the relevant outstanding issues, which will include both procedural and/or substantive approaches, including a potential comprehensive settlement, subject to the parties’ agreement. The Facilitator shall be provided with full access to the records of the clarification process undertaken by the current DSB Chair. The Facilitator-assisted consultations shall remain of a confidential nature.¹²¹

Both members stated their intention to participate in the process in good faith and to ‘seek to avoid taking new measures that may militate against further progress or resolution of the issues between them’.¹²² At the same time, they agreed that participation would not prejudice their rights under the WTO Agreements and reserved their respective rights to terminate the process at any time.¹²³

As further agreed by the parties, the facilitator, Ambassador George Mina of Australia, issued a report on 31 March 2021, which indicated some progress had been made.¹²⁴ The facilitator then issued a second report on 30 June 2022, informing the DSB that the parties had signed on 7 June 2022, an ‘Understanding on Agreed Procedures towards a Comprehensive Settlement of the Dispute in Thailand–Customs and Fiscal Measures on Cigarettes from the Philippines’. By way of the Understanding, the parties ‘agreed to establish a bilateral consultative mechanism (BCM) which will serve as a channel for their respective relevant authorities to cooperate and dialogue on a regular basis, with the objective to build further confidence that will support efforts to reach a comprehensive settlement of their dispute in [Thailand–Cigarettes]’.¹²⁵

5.3 Observations

Thailand–Cigarettes presents facilitation as an alternative path for resolving disputes in the absence of the AB. But it also suggests several reasons to doubt whether that alternative is scalable or desirable.

First, despite the parties' and facilitator's celebration of the success of the process,¹²⁶ the process did not in fact result in a settlement of the parties' dispute. Instead, it resulted in an agreement of the parties to adopt another mechanism (a bilateral consultative mechanism (or BCM)) to ‘serve as a channel for their respective relevant authorities to cooperate and dialogue on a regular basis, with the objective to build further confidence that will support efforts to reach a comprehensive settlement of their dispute in DS 371 … ’.¹²⁷

Second, even the parties' capacity to achieve that limited outcome appears to have been dependent on the parties' close relationship. In a joint communication attached to the second report of the facilitator, the parties emphasized their shared ‘ASEAN treaties and values’ and ‘their close cooperation … consistent with the strong spirit of ASEAN solidarity and friendship which unites them and their commitment to the WTO's rules-based dispute settlement system’.¹²⁸ Where members do not share such common values or close cooperation, facilitation may prove less workable.

Third, both the facilitation process and any further negotiations under the BCM were (are) likely to be affected by bargaining power in a way that compulsory third-party adjudication before the AB (or an arbitration panel) is not. In Thailand–Cigarettes, Thailand's position was that, according to the sequencing agreement entered between the parties, the Philippines remained unauthorized to suspend concessions as long as the compliance reports were on appeal.¹²⁹ The Philippines contested this interpretation, but, at the time of facilitation, it had not yet received authorization to suspend concessions and it was unclear if it would. Assuming Thailand's position was correct, the negotiations (as compared to impartial adjudication) would have favored it over the Philippines. That is because Thailand's apparent alternative to a negotiated solution was the ex ante status quo in which the adverse compliance reports issued against it remained on appeal into the void, and the Philippines remained unauthorized to suspend concessions. By contrast, the Philippines' alternative to a negotiated solution was the continued suffering of Thailand's trade violations with no legal recourse. That dynamic, in which negotiations or facilitation favor the respondent, will reproduce in other disputes as long as appealing into the void remains an option.

Facilitation would be fairer if it was undertaken in the context of a functioning AB, which could impartially decide disputes in the event the parties were unable to reach an agreement. Alternatively, facilitation would be fairer in the context of a broader agreement of the parties to arbitrate any appeals (or to otherwise forego appeals into the void). For the same reasons discussed above with respect to the MPIA, that broader agreement would be easier to achieve before a dispute crystallizes and the parties' interests in effective dispute settlement diverge. Otherwise, agreements to appoint a facilitator will at best paper-over the current dysfunction of WTO dispute settlement.¹³⁰

6. Conclusion

From 2020 to 2022, panels have issued twenty-five reports. Eighteen of those (72%) have been appealed into the void to the now defunct AB. Five of the remaining seven disputes were subject to agreements among the parties (either through the MPIA or bilaterally) to arbitrate their appeals. As discussed above, two of those disputes were in fact appealed and resulted in arbitration awards, while the parties forewent arbitration in the other three.

The numbers are small, but we can nevertheless begin to see some tentative patterns emerging as to how WTO members are navigating the present crisis. First, there has been little self-restraint exhibited by WTO members not bound by the MPIA. Of the twenty-five panel reports issued, twenty-two were issued in disputes where appeals were not subject to the mandatory arbitration agreement of the MPIA. Parties filed appeals into the void in eighteen of those disputes. In other words, where appealing into the void was an option, one of the parties took advantage of that opportunity approximately 82% of the time.

Second, respondents were far more likely to appeal into the void than complainants. Of the eighteen appeals into the void, seventeen were filed by respondents, a phenomenon which can be explained by the asymmetrical benefits that the void has for respondents. Respondents were also more likely to appeal into the void when it was an option than to appeal to arbitration (when arbitration is the exclusive avenue for appeals). As discussed above, respondents appealed to arbitration in only two of the five disputes where arbitration was the compulsory appeals mechanism.

The above observations underscore the importance that parties consent to arbitration before disputes crystallize and the respective postures of the members (complainant or respondent) become clear. That is the method of the MPIA. Nevertheless, as EU–Steel Safeguards and Turkey–Pharmaceutical Products indicate, where parties are engaged in parallel disputes and absorb parallel losses as respondents and parallel victories as complainants, each party will have an interest to preserve its own victory against an appeal into the void by the other party. That can create an opportunity for the parties to agree to parallel arbitration agreements ex post, in which each party promises to refrain from appealing their own loss into the void in exchange for the other party doing the same. That opportunity, however, carries its own perils. It could lead complainants to act inconsistently with their own obligations in order to have something to trade with the other party.

Thailand–Cigarettes demonstrates that members continue to seek out other innovative ways to resolve disputes in light of the AB crisis. The facilitator approach in that dispute led to limited success in the way of the establishment of a bilateral consultative mechanism for continued dialogue and negotiation. However, even that limited success may be hard to reproduce outside of disputes where the parties enjoy a close relationship, as the parties in that dispute did. Moreover, in Thailand–Cigarettes, a facilitator was appointed only after the respondent had appealed two compliance reports into the void. That fact underscores that, absent a functioning appeals mechanism, negotiations, even with a facilitator, will favor respondents whose alternative to a negotiated solution involves bearing no legal consequences for continued trade violations. Facilitation may be a complement to, but not a substitute for, a working appeals mechanism.

From the two appeals resolved by arbitration (Colombia–Frozen Fries and Turkey–Pharmaceutical Products), we can also draw some (very tentative) observations about how adjudicators are navigating the current crisis. While each of those disputes involved different facts and different agreements, there are some commonalities as to how the adjudicators in each approached their roles. Procedurally, each arbitral panel decided their respective appeal within 90 days, and accepted the facts as presented by the panel report. Substantively, there were additional commonalities. First, the arbitrators in both arbitrations departed from past AB precedent. As discussed above, the arbitrators in Colombia–Frozen Fries forged a new path in its interpretation of Article 17.6(ii) of the ADA, while in Turkey–Pharmaceutical Products, the arbitrators did the same with respect to GATT Article III:8. Second, the departures from AB precedent favored granting greater deference to the exercise of national authority. In Colombia–Frozen Fries, the arbitrators’ interpretation granted greater deference to the national authority's interpretation of the ADA. And in Turkey–Pharmaceutical Products, the arbitrators' interpretation expanded, if only slightly, the governmental procurement exception to the National Treatment discipline. Finally, those departures notwithstanding, the arbitrators’ deferential interpretations had no impact on the outcome of the disputes before them. As a practical matter, both disputes ended up in the same place that they would have been had the arbitrators simply affirmed the panel interpretations before them.

We cannot know for sure what motivated the arbitrators' approach in these disputes. However, among many explanations, a hopeful one emerges: As member states muddle through the murky waters of appeals after the Appellate Body, adjudicators may be trying to signal a way out of the swamp.