The WTO’s legal framework for RTAs in
pursuit of a Global Free Trade Network
Student: Yenny Mak
Supervisor: Geraldo Vidigal
Masters: International Trade and Investment Law Version: Final version
Date: January 15, 2018 Wordcount: 12.389
Abstract
This paper answers the question how existing WTO law interacts with the concept of a Global Free Trade Network (GFTN), defined as the situation where any and all pairs of states in the world have concluded a RTA eliminating tariffs. Most previous studies have been economic and inconclusive on whether and how the GFTN will develop. A simple network analysis of the current network of 282 RTAs in the WTO’s online database, and 175 states and customs territories globally, provides further statistical insight in RTA proliferation. By the end of 2017, only 8.4 percent of the bilateral free trade relationships necessary for establishment of the GFTN have been established. None of the models developed in the literature discussed in this paper encompass all phenomena and trends of RTA proliferation found in this current network. While economic modelling seems to be unable to accurately predict RTA formation in light of the GFTN, Article XXIV of the GATT and paragraph 2(c) of the Enabling Clause may be able to guide regionalism towards the GFTN. If ‘substantially all the trade’ is read strictly, meaning RTA concessions must be symmetrical, and eliminate tariffs for a high percentage of tariff lines and actual trade flows, the internal requirement in Article XXIV:8(a)(i) and 8(b) of the GATT stimulates the establishment of the GFTN. On the other hand, ‘substantially the same duties’ in Article XXIV:8(a)(ii) concerning the external requirement for CUs, should be interpreted flexibly. If not, it would cause CUs to conflict with other RTAs. Resolving such conflict among RTAs does not resolve the violation of the external requirement. As such the external requirement forms a stumbling block to the GFTN, and is obsolete once the GFTN is established. Similarly, the economic test of Article XXIV:5(a) and 5(b) is obsolete, and becomes increasingly so as the RTA network grows. Interaction between RTAs in a hub-and-spokes system provides a similar protection to third countries. As the RTA network grows, the number of third countries protected by the economic test decreases. The abolishment of the external requirement and economic test may therefore be considered. This conclusion cannot be drawn for the authorization for PSAs in paragraph 2(c) of the Enabling Clause. The more flexible PSAs may form stumbling blocks for FTAs and CUs that make up the GFTN, but may also be seen as literal building blocks of FTAs and CUS in the same sense FTAs and CUs build the GFTN.
Contents
Chapter 1 Introduction ... 4
Chapter 2 The economic process of the GFTN ... 7
2.1 Literature Review ... 8
2.2 Current Network ... 12
2.2.1 Dataset ... 13
2.2.2 Results and discussion ... 15
Chapter 3 The legal road to the GFTN ... 17
3.1 Internal Requirement ... 19 3.1.1 Qualitative aspect ... 20 3.1.2 Quantitative aspect ... 21 3.2 External requirement ... 25 3.3 Economic Test ... 28 3.4 Enabling Clause ... 29 Chapter 4 Conclusion ... 30 Bibliography ... 32 Literature ... 32
GATT/WTO agreements, documents, and case law ... 35
National law and international treaties ... 36
Chapter 1 Introduction
FIGURE 1:PROLIFERATION OF RTAS OVER TIME
Source: World Trade Organization (WTO), ‘Regional Trade Agreements: Facts and Figures’ <https://www.wto.org/english/tratop_e/region_e/regfac_e.htm> accessed 20 October 2017.
As of December 18th, 2017 the World Trade Organization’s (WTO) database, the Regional Trade Agreement Information System (RTA-IS) contains 283 regional trade agreements (RTA) on goods in force.1 Figure 1 shows how the number of RTAs have proliferated since 1948. This trend has raised concerns about the relationship between regionalism and multilateralism, resulting in the stumbling block versus building block debate.2 This paper contributes to the discussion with a legal analysis of the concept of a Global Free Trade Network (GFTN). This refers to the situation that the accumulation of RTAs allows goods to be traded freely on a global level.
The concept of the GFTN is based on the presumption that proliferation perseveres, so that eventually each country will have established a RTA-relationship with each other country in
1 WTO, “Regional Trade Agreement Information System (RTA-IS)” <rtais.wto.org> accessed 28 December
2017.
2 A Panagariya, ‘Preferential Trade Liberalization: The Traditional Theory and New Developments’ (2000) 38
the world. In terms of social network analysis, this forms a ‘complete network’.3 Support for
this can be found in figure 1. The number of active RTAs, thus in force, has increased consistently especially after 1991. The decline in 2004 and 2007 can be ignored as it reflects a consolidation of trade relations. While physical RTAs were terminated, the EU’s RTAs actually expanded with accessions consequent to the enlargements of the European Union (EU).4 Even now new RTAs are being negotiated, such as between China and Mauritius, which launched on December 12th, 2017.5
Currently, the RTA-IS contains three types of RTAs on goods: 17 agreements establish Customs Unions (CU), 245 agreements for Free Trade Areas (FTA), and 21 Partial Scope Agreements (PSA). With regards to CUs and FTAs, the RTA-IS follows the definitions found in Article XXIV of the General Agreement on Tariffs and Trade (GATT).6 As such constituent parties eliminate tariffs on substantially all trade between them, and in case of CUs also establish substantially the same external trade policy.7 Additionally, PSAs are defined in the RTA-IS as RTAs under the Enabling Clause, that cover only certain products.8 Paragraph 2(c) of the Enabling Clause allows for such agreements among developing if they reduce tariffs.9 The provision does provide a threshold for product coverage. The interaction between the WTO law provisions for RTAs and the GFTN forms the topic of this paper. The research question to be answered is: How do the GFTN and WTO’s legal framework for RTAs
interact?
Article XXIV:4 of the GATT already points out that RTAs are authorized to increase the possibilities to liberalize trade. Therefore, the purpose of RTAs should be to break down trade barriers between constituent parties without raising trade barriers towards third countries. In this purpose lies the objective of the GFTN as final stage of RTA proliferation, where all trade barriers are eliminated. Among others, Low (2014) had suggested that the WTO should adopt
3 W de Nooy, A Mrvar and V Batagelj, Exploratory Social Network Analysis with Pajek (CUP, Cambridge
2011), 73.
4 Notification from Turkey (7 May 2004) WT/REG/GEN/N/2; Notification from the European Communities and
its Member States to the Committee on Regional Trade Agreements (CRTA) (5 October 2004) WT/REG/GEN/N/3; Notification from the European Communities and its Member States to the CRTA (2 March 2007) WT/REG/GEN/N/4; Notification from the CEFTA Member States to the CRTA (7 August 2007) WT/REG/GEN/N/5.
5 --, ‘China and Mauritius Officially Launch the FTA Negotiations’ (China FTA Network, 14 December 2017)
<http://fta.mofcom.gov.cn/enarticle/enrelease/201712/36682_1.html> accessed 28 December 2017.
6 WTO, “RTA-IS User Guide” <http://bit.ly/2lzYOEd> accessed 28 August 2017 (hereinafter “RTA-IS User
Guide”) [I.C].
7 The General Agreement on Tariffs and Trade (15 April 1994), 1869 U.N.T.S. 190 (hereinafter “GATT”), art.
XXIV:8(a) and 8(b).
8 RTA-IS User Guide (n 6) [I.C].
9 Decision on Differential and More Favourable Treatment, Reciprocity and Fuller Participation of Developing
Countries (Decision of 28 November 1979) GATT Document L/4903, BISD 26S/203 (hereinafter “Enabling Clause”)
‘soft law’ to help reformulate the legal framework to align regionalism with global free trade. He actually rejected the possibility that such alignment can be achieved by multilateralizing regionalism in the sense of expanding and fusing trade blocs into a complete network, because he deemed it a self-executing process based on economic interests that could not be guided.10 The goal of this paper is to explore whether the WTO’s legal framework for RTAs can provide such guidance, if the WTO law provisions are interpreted from the perspective that RTAs are incremental building blocks of the GFTN. It should be mentioned that this paper focuses solely on tariffs on goods. Services, rules of origin, and other provisions of RTAs are outside the scope of this paper. Furthermore, since the GFTN is not an existing situation the analysis in this paper has a hypothetical nature.
Two sub-questions help answer the main research question: What is the current state of affairs of RTAs? How can the GFTN as the final stage of RTA proliferation be established? As Charnovitz stated: “we should try to discern the horizon”.11 Thus, chapter 2 aims to describe the GFTN more concretely by addressing the two questions. Like Low (2014) had pointed out, the GFTN is mostly an economic concept, so a concrete understanding of the GFTN is unavoidably in economic terms. Paragraph 2.1 will first review the existing literature, and paragraph 2.2 performs a simple social network analysis on the current network of RTAs in force. By creating a better understanding of the GFTN, chapter 2 lays the groundwork for the legal analysis in chapter 3.
Chapter 3 addresses aforementioned sub-questions from a legal and normative perspective. The sub-questions can be rephrased into: how are Article XXIV of the GATT and the Enabling Clause interpreted, and how does this influence the establishment of the GFTN? The other side of the same coin for the second question is how the prospect and aim towards the GFTN can influence this interpretation. Paragraph 3.1 to 3.3 discusses the three requirements of Article XXIV of the GATT: the internal requirement for CUs and FTAs to eliminate tariffs on substantially all trade of Article XXIV:8(a)(i) respectively 8(b) of the GATT; the external requirement for CUs to establish substantially the same trade policy towards third countries in sub-paragraph 8(a)(ii), and; the economic test comparing trade restrictions towards third countries before and after the formation of the CU or FTA sub-paragraphs 5(a) respectively 5(b). Paragraph 3.4 discusses aforementioned questions with regards to the Enabling Clause. Chapter 4 concludes this paper.
10 P Low, ‘Preferentialism in Trade Relations: Challenges for the World Trade Organization’ (2014) Asian
Development Bank Institute (ADBI) Working Paper Series No. 478, 12-14.
11 S Charnovitz, ‘The World Trade Organization in 2020’ (2004) 1 Journal of International Law and
Chapter 2 The economic process of the GFTN
In the history of RTAs, three waves of regionalism can be identified. The first wave started in 1958 with the formation of the European Community (EC), now the EU. The second wave started in the 1980s, fuelled by the United States’ (US) new pro-RTA policy.12 Finally, with
the conclusion of the Uruguay Round in 1994 came the third wave,13 marked by the advent of Asia.14 Regionalism can be defined as any trade agreement involving two or more parties, but less than all WTO members.15 From a global perspective, regionalism refers to a trade agreement including less than all existing countries. If such agreements reduce barriers to trade between signatories, they establish trade blocs.16 Thus, CUs and FTAs are trade blocs, but so are the associations established by PSAs.
The possibility to form CUs and FTAs has been included in the GATT since 1947, as exceptions to the cornerstone Most Favoured Nation-principle (MFN), which requires tariffs to be applied in a non-discriminatory manner.17 The rationale behind such exception stems
from the history of CUs as one of the stages in political unifications to form single nations.18
Since constituent parties would fully integrate on trade, there was no good reason not to allow CUs.19 The authorization of FTAs was included for the situation that CUs were not politically
feasible due to the burden of harmonizing external policy. The belief was that FTAs would eventually be replaced by CUs.20 To accommodate developing countries and least-developed countries (LDC), the possibility of PSAs was added in the Enabling Clause in 1979.21
Although RTAs have been authorized in the multilateral framework of the WTO, their effect on world trade and the multilateral system has been much debated. The idea of the GFTN was conceived in this debate. The literature on this issue is too vast to review exhaustively, but paragraph 2.1 provides a brief and selective overview. Some of the studies develop economic models to analyse the effects and probability of RTA formation. The diversity in models and
12 J Bhagwati, ‘Regionalism and multilateralism: an overview’ in J de Melo and A Panagariya (eds), New
Dimensions in Regional Integration (CUP, Cambridge 1993), 28-30.
13 T Carpenter, ‘A Historical Perspective on Regionalism’ in R Baldwin and P Low, Multilateralizing
Regionalism (CUP, Cambridge 2008), 17-23.
14 WTO, World Trade Report 2011. The WTO and preferential trade agreements: From co-existence to
coherence (Geneva, WTO 2011) (hereinafter “World Trade Report 2011”) 53.
15 Carpenter 2008 (n 13), 13.
16 JJ Schott, ‘Trading Blocs and the World Trading System’ (1991) 14 The World Economy 1, 1. 17 GATT (n 7), art. I:1.
18 K Chase, ‘Multilateralism Compromised: The Mysterious Origins of GATT Article XXIV’ (2006) 5 World
Trade Review 1, 4.
19 Bhagwati 1993 (n 12), 25-26.
20 Z Hafez, ‘Weak Discipline: GATT Article XXIV and the Emerging WTO Jurisprudence on RTAs’ (2003) 79
North Dakota Law Review 879, 883-884.
21 AA Yusuf, ‘”Differential and more Favourable Treatment”: The GATT Enabling Clause’ (1980) 14 Journal of
ambiguous findings indicate the difficulty of formulating a comprehensive model for the multi-faceted phenomenon of regionalism.22 The extent to which previous research has been
able to grasp real-life trends is evaluated in paragraph 2.2, which compares the current network of RTAs to the findings of previous studies.
2.1 Literature Review
The first issue of regionalism is the effect of trade blocs on world trade, determined by whether the elimination of tariffs within trade blocs shifts demand and supply from high-cost producers to lower-cost producers. If trade shifts to low-cost producers, the effect on world trade is positive thus trade creating. If the shift is to high-cost producers, the effect is negative thus trade diverting. While RTAs were initially viewed as favourable to free trade, Viner (1950) introduced the idea that CUs could also be trade diverting depending on several factors. For example, the level of tariffs imposed by third countries, the level of external tariffs harmonized by the CU, and whether producers in the CU are able to benefit from economies of scale. The effects can vary per product and industry, change over time.23 Usually, RTAs will be trade creating for constituent parties but trade diverting for third countries.24
Despite trade diverting effects, RTAs continued to proliferate and it became more and more clear that regionalism would not disappear from the global trading system. Bhagwati (1993) stressed the importance of guiding RTAs to be trade creating and in harmony with multilateral free trade.25 Besides Viner’s static analysis of RTAs, research also started to focus on the dynamic time-path issue.26 This issue can be divided into three questions. First, will regionalism lead to a global trade bloc? Second, will regionalism help or hinder multilateralism, otherwise known as the stumbling block versus building block debate. Finally, how does multilateralism impact regionalism?27
With regard to the first question, Viner (1950) noted that a global CU would be equal to global free trade,28 which can be seen as the ultimate goal of the GATT and the WTO.29 In
other words, a trade bloc with all countries in the world establishes global free trade. However,
22 See for example J Goto and K Hamada, ‘Regional Economic Integration and Article XXIV of the GATT’
(1999) 7 Review of International Economics 555, 556.
23 J Viner, The customs union issue (New York: Carnegie Endowment for International Peace 1950), 41-45 and
51-52.
24 Bhagwati 1993 (n 12), 33; Goto and Hamada 1999 (n 22), 568. 25 Bhagwati 1993 (n 12), 46.
26 Panagariya 2000 (n 2), 287. 27 Panagariya 2000 (n 2), 317-318. 28 Viner 1950 (n 23), 52.
studies show that forming a global trade bloc at once is not feasible.30 The Doha deadlock is
evidence of the complexity of negotiating with many participants. A more viable option is sequential trade liberalization, for which there are two approaches. Either free trade spreads with an expanding regime, meaning RTAs expand through accessions, or with an overlapping regime whereby countries form multiple RTAs in a hub-and-spoke system.31 If either regime continues until all possible countries are included and all trade barriers are eliminated, regionalism is a building block for multilateral trade liberalization. Otherwise, RTAs are stumbling blocks.32
The expanding regime has been supported by Baldwin (2006), who reasoned that an accession would be triggered by an event that emphasizes the benefits of membership to the trade bloc. The discrimination against third countries would be stronger after the accession, strengthening the incentive for other third countries to also accede. One accession leads to another, and eventually all countries will have acceded thereby forming a global trade bloc.33 Yi (1996) likewise found that when CUs allow accessions by any country, small CUs can grow into global CUs.34 On the other hand, Goto and Hamada (1999) found that expansion ceases when the bloc encompasses half the world. Further enlargements would be trade diverting towards constituent parties, so that they would oppose any subsequent accessions.35 The more countries enjoy the preferential benefits from a RTA, the less exclusive the preferences become. The same reasoning applies to an overlapping regime. Tobin and Busch (2010) concluded that developing countries should be selective in choosing RTA-partners. Because of preference erosion, developed countries would be less willing to conclude a RTA with a developing country, if that developing country has concluded nine bilateral RTAs previously.36
Nevertheless Nomura et al. (2013) concluded an overlapping regime would lead to global free trade unlike the expanding regime. If a RTA is formed between two developed countries, a hub-and-spoke with a third developing country is formed through a second RTA, followed by
30 See also DR Trupkin, ‘On the Viability of a Multilateral Trade Agreement: A Political-Economy Approach’
(2010) 27 Journal of Business Strategies 54, 68-70; R Nomura et al., ‘Does a Bilateral FTA Pave the Way for Multilateral Free Trade’ (2013) 21 Review of International Economics 164, 172-173.
31 Nomura et al 2013 (n 30), 164-165.
32 T Furuwawa and H Konishi, ‘Free Trade Networks’ (2007) 72 Journal of International Economics 310, 311. 33 RE Baldwin, ‘Multilateralising Regionalism: Spaghetti Bowls as Building Blocs on the Path to Global Free
Trade’ (2006) 29 The World Economy 1451, 1467-1468.
34 SS Yi, ‘Endogenous Formation of Customs Unions under Imperfect Competition: Open Regionalism is Good’
(1996) 41 Journal of International Economics 153, 155.
35 Goto and Hamada 1999 (n 22), 564.
36 JL Tobin and ML Busch, ‘A Bit is Better Than a Lot: Bilateral Investment Treaties and Preferential Trade
a third RTA between the two spokes. In this manner, the overlapping regime leads to a complete network of RTAs.37 According to Mukunoki and Tachi (2006), this is more likely
with FTAs, because CUs limit the flexibility of constituent parties to conclude subsequent RTAs with third countries.38 Goyal and Joshi (2006) also remarked that countries cannot join multiple CUs.39 Lee and Bai (2013) performed a social network analysis of FTAs and also that the FTA partner of a FTA partner is likely to be an FTA partner. In other words, two spokes are likely to have concluded a FTA. At the same time, FTAs should not be thought of as bilateral, because trilateral relation are more stable.40
Furthermore, Furusawa and Konishi (2007) found the overlapping regime is less likely to occur with countries of symmetric market and industry sizes, if either of them is already party to a CU, compared to when either of them is party to a FTA. For asymmetric countries a pre-existing CU can be beneficial as it may average out asymmetries, which is important because a FTA will only be formed between symmetric countries given that products are independent. This in turn implies developing and developed countries will form mutually exclusive trade blocs.41 This finding is also backed by Yi (1996), who found that the expanding regime would lead to two mutually exclusive CUs, if accession to the CU requires consent of all existing constituent parties. Daisaka and Furusawa (2014) similarly concluded the world will fragment into several trade blocs. A complete network of RTAs is also possible in rare cases, but in most scenarios they observe a near-complete network that is still able to attain global free trade.42
As mentioned earlier, unless regionalism leads to a complete network or global trade bloc, it is deemed a stumbling block. However, Saggi and Yildiz (2011) stated that RTAs can also be partial building blocks. When multilateral negotiations are unfruitful, regionalism results in at least partial trade liberalization. The possibility to pursue RTAs can even stimulate countries to pursue multilateral trade liberalization.43 Bhagwati (1993) also considered regionalism as supplement to multilateral negotiations. RTAs allow deeper integration between willing
37 Nomura et al. 2013 (n 30), 170-172.
38 H Mukunoki and K Tachi, ‘Multilateralism and Hub-and-Spoke Bilateralism’ (2006) 14 review of
International Economics 658, 670.
39 S Goyal and S Joshi, ‘Bilateralism and Free Trade’ (2006) 47 International Economic Review 749, 751. 40 T Lee and BI Bai, ‘Network Analysis of Free Trade Agreements. Homophily and Transitivity’ (2013) 11 The
Korean Journal of International Studies 263, 288
41 Furusawa and Konishi 2007 (n 32), 314.
42 H Daisaka and T Furusawa, ‘Dynamic Free Trade Networks: Some Numerical Results’ (2014) 22 Review of
International Economics 469, 485.
43 K Saggi and HM Yildiz, ‘Bilateral Free Trade Agreements and the Feasibility of Multilateral Free Trade’
members at a faster pace than multilateral trade liberalization.44 This function is also
perceived in relation to other RTAs. For example, China and Singapore concluded a FTA in 2008 to establish further integration than the FTA of 2005 between China and the Association of Southeast Asian Nations (ASEAN).45 According to MacMillan (2014) regionalism is further stimulated by the expanding WTO membership and process of consensus decision-making in the WTO.46 These statements argue against expelling regionalism from the WTO in its entirety just because it does not lead to the GFTN. The debate can be specified further into the question: is some liberalization not better than none? Is the formation of major trade groupings not better than a completely fragmented global trade? The answer to these questions cannot be answered easily. As Viner (1950) pointed out, a distinction in the degree of preference can be illogical. It does not make sense to allow preferential treatment at 100 percent, but to classify it as maximum evil at 99 percent.47 Similarly, it does not make sense to write off regionalism because the resulting network is 99 percent complete, but not 100. Furthermore, Baldwin (2006) argued that a trade diverting RTA works as stumbling block, because it lowers the amount of trade given a certain MFN tariff, thereby raising the optimal MFN tariff.48 This statement may be true if MFN tariffs are set at optimal levels, but it does not apply to the current situation of world trade since the WTO only reduces MFN tariffs and never increases them.49 Thus, it is not convincing to say regionalism causes multilateral trade
liberalization to reverse. At worst, multilateral trade liberalization is slowed or halted. In that regard, the findings of previous research is ambiguous. Liao (2007) concluded that RTAs would increase the cost of MFN reductions,50 whereas Ornelas (2005) found that FTAs
induced constituent parties to lower external tariffs.51
44 Bhagwati 1993 (n 12), 26.
45 J Pauwelyn and W Alschner, ‘Forget About the WTO: The Network of Relations between PTAs and “Double
PTAs”’ in A Dür and M Elsig, Trade Cooperation. The purpose, Design and Effects of Preferential Trade Agreements (CUP, Cambridge 2015), 25.
46 E MacMillan, ‘Explaining rising regionalism and failing multilateralism: consensus decision-making and
expanding WTO membership’ (2014) 11 International Economics and economic Policy 599.
47 Viner 1950 (n 23), 49-50. 48 Baldwin 2006 (n 33), 1471.
49 Agreement Establishing the World Trade Organization (Signed on 15 April 1994), 1867 U.N.T.S. 154
(hereinafter “Marrakesh Agreement”) Preamble; Ministerial Declaration (Adopted on 14 November 2001) WT/MIN(01)/DEC/1 [16]; Also see Goto and Hamada 1999 (n 21), 556.
50 N Limao, ‘Are Preferential Trade Agreements with Non-trade Objectives a Stumbling Block for Multilateral
Liberalization’ (2007) 74 Review of Economic Studies 821, 823-824.
51 E Ornelas, ‘Endogenous Free Trade Agreements and the Multilateral Trading System’ (2005) 67 Journal of
2.2 Current Network
The ambiguous findings summed in paragraph 2.1 only partially shows the complexity of the issue of regionalism. If one were to dive into the contents of RTAs, the lack of coherence in tariffs and rules of origin of different RTAs show a chaos, that has sometimes been referred to as a ‘spaghetti bowl’.52 Social network analysis makes it possible to identify patterns and gain
factual insight into such phenomena.53 Pauwelyn and Alschner (2015) used social network analysis on 248 RTAs to discuss the issue of ‘double PTAs’, or ‘double RTAs’, which is the phenomenon where two countries have concluded multiple RTAs.54 Social network analysis investigates a system of social relationships, or ‘ties’, between entities, or ‘actors’.55 Ties can
be formed between two actors, which means that RTAs with more than two constituent parties will be approached as a compilation of bilateral ties. In terms of social network analysis, plurilateral RTAs form ‘complete subnetworks’ between constituent parties.56
A simple analysis is performed with data from the RTA-IS. It is estimated that over 100 RTAs in force among developing countries remain non-notified.57 Since the RTA-IS depends on notifications and announcements by WTO members,58 the non-notified agreements are not included in the analysis of paragraph 2.2.
A full briefing of the contents of the nearly 300 RTAs in the RTA-IS goes beyond the limitations of this paper. For the sake of the analysis in this paragraph, the precise content of RTAs is therefore disregarded. More specifically, it is assumed that RTAs cover all trade. The flexibility of the requirement of ‘substantially all trade’ in Article XXIV:8(a)(i) respectively 8(b) of the GATT will be discussed in detail in paragraph 3.1, but is put aside for now. With regard to PSAs, the issue is somewhat more complicated, because the RTA-IS defines PSAs in terms of product coverage.59 If a RTA covers substantially all trade yet merely reduces tariffs, it would be authorized under the Enabling Clause, but it is neither a PSA nor a CU or a FTA pursuant to the RTA-IS. One explanation may be that such agreements have not been concluded or notified to the WTO. The discrepancy between the RTA-IS and the Enabling Clause is accounted for in this paper by assuming that the RTA-IS meant to provide a closed typology for RTAs. Regardless of the definition in the RTA-IS, PSAs are therefore deemed to
52 J Bhagwati, D Greenaway and A Panagariya, ‘Trading Preferentially: Theory and Policy’ (1998) 108 The
Economic Journal 1128, 1139.
53 Pauwelyn and Alschner 2015 (n 45), 502.
54 Pauwelyn and Alschner (2015) use the term PTAs to refer to RTAs. 55 De Nooy, Mrvar and Batagelj 2011 (n 3), 1, 5-6.
56 De Nooy, Mrvar and Batagelj 2011 (n 3), 73. 57 World Trade Report 2011 (n 14), 54. 58 RTA-IS User Guide (n 6) [II]. 59 RTA-IS User Guide (n 6) [II.C].
be RTAs that do not comply with Article XXIV of the GATT, but comply with paragraph 2(c) of the Enabling Clause. Since product coverage of RTAs is ignored for the sake of analysis, PSAs are distinguished from CUs and FTAs by the degree of trade liberalization. While CUs and FTAs eliminate tariffs, PSAs establish lower but nonzero tariffs.
The precise degree of tariff reduction by PSAs will not be discussed, but it is assumed that signatories to a PSA concede to the same degree of reduction. Consequently, the AB-tie that is established by a PSA between A and B is equal to the BA-tie. The same applies to CUs and FTAs, meaning RTAs create mutual and symmetric bilateral ties.
The GFTN requires free trade to be established between all countries in the world. Under aforementioned assumptions, once free trade is established between A and B, any subsequent RTA between A and B is redundant in the GFTN. Therefore, the GFTN is a ‘simple network’ made of unique symmetric bilateral ties.60 The number of possible ties given n actors can be calculated with the formula 𝑛 ×(𝑛−1)
2 .
61 Moreover, since PSAs liberalize trade but do not
establish ‘free trade’, the GFTN only consists of CUs and FTAs. Paragraph 2.2.1 provides the criteria which have been used to select the data and paragraph 2.2.2 discusses the results.
2.2.1 Dataset
The RTA-IS contains 283 active and notified RTAs, 18 notified accessions, as well as 2 active RTAs for which early announcements were made. Accessions are taken into account in the analysis, but are not counted as separate agreements. One of the announced RTAs will be included, namely the FTA between the European Free Trade Association (EFTA) and the Gulf Cooperation Council (GCC), which had entered into force on July 1st, 2014. The other early announcement, the Cross-Straits Economic Cooperation Framework Agreement, is excluded because it is not reported whether it concerns a CU, FTA, or PSA. PSAs are included in the analysis to provide insight into the broader context of preferential trade liberalization besides the focus on free trade. For this reason the Global system of Trade Preferences among Developing countries (GSTP) and the Protocol on Trade Negotiations (PTN) are also included.62
Besides RTAs the dataset is also selected on actors. Actors must be able to conclude RTAs in order to participate in the GFTN. Hence an actor must be sovereign or autonomous over
60 De Nooy, Mrvar and Batagelj 2005 (n 3), p. 7-8. 61 See Pauwelyn and Alschner 2015 (n 45), 504. 62 Cf. Pauwelyn and Alschner 2015 (n 45), footnote 14.
external trade matters, which are also the criteria for WTO membership.63 In other words,
actors are WTO members and those eligible for WTO membership.
As of 2016 the WTO has 164 members.64 Although RTAs are not deemed actors in this paper,
the EU forms an exception as it has acceded to the WTO.65 Considering the EU has exclusive competence over trade matters, so RTAs with EU member states are concluded with the EU as whole,66 the EU and its 28 member states are seen as one collective actor in this paper.67 Additionally, the WTO has 23 observer governments. With exception of the Holy See, the observers are pursuing WTO membership.68
Furthermore, the fact that a country or customs territory is party to a RTA reported in the RTA-IS is sufficient to see it as an actor regardless of WTO membership. Based on 283 RTAs, including the EFTA-GCC FTA and excluding the EC Treaty, 33 non-WTO RTA parties are identified. This includes the EU’s 19 Overseas Countries and Territories (OCT), but due to the uncertainty about their full autonomy over trade matters they are excluded from the dataset anyway. For example, Montserrat is party to the Caribbean Community and Common Market (CARICOM), while RTAs with the countries of the Netherlands Antilles are formally concluded by the Kingdom of the Netherlands.69 A full understanding of the special relationship between EU and OCTs would require extensive investigation beyond the scope of this paper, so any OCT is simply not recognized as actor. As a result the EU-OCT Association is excluded from the dataset, and so are the relationships between Montserrat and the other CARICOM signatories. Consequently, the dataset, thus the existing network of RTAs, contains 170 actors (136 WTO, 23 WTO Observers, 14 RTA parties) and 282 RTAs (21 PSAs, 16 CUs, 244 FTAs, 1 early announced FTA).
Ideally, the GFTN includes all possible sovereign or autonomous countries or customs territories in the world. In addition to WTO members and reported non-WTO RTA parties, Palau and Monaco are identified as potential actors that are capable of concluding RTAs. They are taken from the United Nations (UN) membership, which is only open to states,70
63 Marrakesh Agreement (n 49), art XIII:1.
64 WTO, ‘Members and Observers’ <https://www.wto.org/english/thewto_e/whatis_e/tif_e/org6_e.htm>
accessed 3 January 2018 (hereinafter “WTO Members and Observers”).
65 Marrakesh Agreement (n 49), art XI:1.
66 Treaty on the Functioning of the European Union (TFEU) art 207.
67 Cf. Pauwelyn and Alschner 2015 (n 45), 504; World Trade Report 2011 (n 14), 48. 68 WTO Members and Observers (n 63).
69 Statute of the Kingdom of the Netherlands, art 6(1) and 11(3). 70 Charter of the UN (Signed on 26 June 1945) 1 U.N.T.S. XVI, art 4.
which in turn means they are eligible for WTO membership. To conclude, the GFTN would comprise of 175 actors.71
2.2.2 Results and discussion
The 282 RTAs included in the dataset form 3404 bilateral ties if double RTAs are counted separately. If the duplicate ties are counted only once, the existing network has 2209 unique bilateral ties. The number of existing ties decreases further if only free trade is considered. Thus, the 261 CUs and FTAs form 2093 ties including duplicates, and only 1281 unique ties. Compared to the GFTN with 175 actors who can form 15225 unique bilateral ties, the current network has only established 8.4 percent of the preferential free trade that is needed for the GFTN.
The most active actors are the EU (38 RTAs), the EFTA country (28-31 RTAs per country), Chile (29), Turkey and Singapore (both 22). Mauritania, Monaco, Palau, Somalia, Timor-Leste, Sao Tome and Principe have not concluded any RTAs. Excluding these 6 actors, the average activity is 6.1 RTAs per actor. Cuba, Iran, North-Korea, and the Marshall Islands have only concluded PSAs, meaning they have not established free trade with any other country. Based on the 261 CUs and FTAs, the most active actors are the EU (71 unique ties), the EFTA countries (40-43 ties per country), Egypt (39), Sudan and Libya (both 33). Excluding the actors who have not formed any ties or free trade-ties, the average number of unique RTA-ties per actor is 26.3, and the average number of unique free trade-ties is 15.6. These numbers have far exceeded the tipping point of 9 bilateral RTAs determined by Tobin and Busch (2010). A more detailed discussion requires the identification of the developing countries and LDCs. For this, two sources can be used. Firstly, the text of paragraph 2(c) of the Enabling Clause only authorizes RTAs among less-developed countries, referring to developing countries and LDCs. Assuming the RTAs in the RTA-IS comply with the legal requirements set by the WTO, it follows that all signatories to RTAs notified under the Enabling Clause are developing countries or LDCs. In the dataset, 44 RTAs are notified under the Enabling Clause, which allows for the identification of 113 actors as less-developed. Secondly, the status of LDCs is determined by the UN,72 and the UN has labelled 47 countries accordingly.73 Eliminating the duplicates, there are a total of 123 less-developed countries.74
71 Cf. Carpenter 2009 (n 13), 23 who concludes 195 state actors were eligible for WTO membership in 2008. 72 UN, ‘Inclusion in the LDC category’ <http://bit.ly/2qPMENF> accessed 23 December 2017.
Of the less-developed countries, 11 countries have more than 9 unique free trade-ties with developed countries. For example, Chile has formed 28 free trade-ties, of which 12 are with developed countries.
Furthermore, as Lee and Bai (2013) suggested, the conventional thinking of RTAs as bilateral agreements is outdated. Many RTAs (30.4 percent) are plurilateral with 3 or more actors. The largest trade blocs are the GSTP (42 actors), the Common Market for Eastern and Southern Africa (COMESA) with 19 actors and the Pan-Arab FTA (PAFTA) with 18 actors. Both the PAFTA and the COMESA expanded via accessions. Still, the original membership of the COMESA before Egypt’s accession counted 18 actors. These observations can be compared with some of the findings of Nomura et al. (2013). First, the fact that accessions have occurred, especially the EC enlargements, refutes the finding that the expanding regime will not occur. Second, the relatively large membership with which the COMESA was founded can be interpreted in light of the finding that multilateral agreements will not be concluded in one round of negotiations in two ways. On the one hand, Nomura et al. had developed a model for a setting of 3 actors and concluded a trilateral agreement was not feasible. Clearly, the COMESA is a plurilateral agreement that goes beyond the trilateral agreement Nomura et al. had written off. On the other hand, the COMESA only included 18 of the 174 actors in the world, so it does not prove a multilateral agreement is feasible.
Additionally, some of the ties are formed by RTAs between trade blocs, for example the EFTA has concluded a FTA with the Southern African Customs Union (SACU). While the EFTA countries are developed, the SACU countries are not. Therefore, the EFTA-SACU FTA bridges the gap between developing and developed countries which Furusawa and Konishi (2007) predicted. In other words, the EFTA-SACU FTA counters the statement that the world will develop into two mutually exclusive trade blocs divided by level of development. In fact, the gap between plurilateral RTAs is often overcome. For example, the Southern Common Market (MERCOSUR) has concluded a PSA with SACU, and the EU has formed a FTA with the Southern African Development Community (SADC). Also, Oman and Qatar are both party to the PAFTA and the GCC, and Egypt, Libya, and Sudan are party to the PAFTA as well as the COMESA. Swaziland is even party to two CUs, namely the COMESA and the SACU. Thus, the gap can also be bridged by common membership.
74 This method is not flawless. For example, the South Pacific Regional Trade and Economic Cooperation
Agreement is notified under the Enabling Clause. Australia is party to this agreement, but it can be seriously doubted that Australia is a developing country.
Chapter 3 The legal road to the GFTN
Chapter 2 discussed the economic process of forming the GFTN and the doubt about its attainability. The majority of previous research concluded the GFTN is possible yet not probable. The analysis of the current network refutes most of the findings that underlie this conclusion. From an optimistic perspective, this can be taken as support for the assumption that regionalism will lead to the GFTN, although the current network is a long way from complete.
Chapter 3 now turns to evaluate the interaction between the economic concept of the GFTN and WTO law. For this, it is necessary to answer how the WTO law provisions are interpreted commonly, and how they can be interpreted from the teleological perspective of the GFTN. The interpretation of the legal requirements was avoided in paragraph 2.2 by assuming that the RTAs comply with WTO law. Yet it is unknown whether this assumption holds.
First of all, the WTO provisions on the formation of RTAs are not actually enforced.75
Initially, notified RTAs were examined by ad hoc working parties, and since 1996 by the Committee on Regional Trade Agreements (CRTA).76 However, with the adoption of the Transparency Mechanism for RTAs in 2006 the examinations were replaced by considerations.77 A notified RTA is now considered by the other WTO members and only scrutinized if the RTA’s compliance with WTO law is challenged before a WTO panel.78
Since the establishment of the WTO’s Dispute Settlement Body (DSB) this has rarely happened. The relevant cases can be counted on one hand: Turkey – Textiles, Brazil – Tyres,
Peru – Agricultural Products.79 Nevertheless, the Panel in Brazil – Tyres avoided interpreting Article XXIV by exercising judicial economy.80 Even though the Appellate Body disagreed with the Panel’s conclusion, it did not interpret Article XXIV itself.81
75 WTO, The Future of the WTO: Addressing institutional challenges in the new millennium (Report by the
Consultative Board to the Director-General Supachai Panitchpakdi) (2004) <https://www.wto.org/english/thewto_e/10anniv_e/future_wto_e.pdf> accessed 20 September 2017 [75].
76 R Wolfrum, PT Stoll and K Kaiser, WTO: Institutions and Dispute Settlement (Martinus Nijhoff Publishers,
Leiden 2006), 250; CRTA (Decision of 6 February 1996) WT/L/127 [1(a)].
77 Transparency Mechanism for Regional Trade Agreements (Decision of 14 December 2006) WT/L/671
(hereinafter “TM RTA”).
78 TM RTA (n 74) [4].
79 Panel Report, Turkey – Restrictions on Imports of Textile and Clothing Products (adopted 19 November 1999)
WT/DS34/R (hereinafter “PR, Turkey – Textiles”); Panel Report, Brazil – Measures Affecting Imports of Retreaded Tires (adopted 17 December 2007) WT/DS332/R (hereinafter “PR, Brazil - Tires”); Panel Report, Peru – Additional Duty on Imports of Certain Agricultural Products (adopted 31 July 2015) WT/DS457/R.
80 PR, Brazil – Tires (n 79) [7.456].
81 Appellate Body Report, Brazil – Measures Affecting Imports of Retreaded Tires (adopted 17 December 2007)
Second, it is not clear how the legal requirements should be interpreted. This lack of clarity is also one of the reasons why only one of the examined RTAs has been found fully consistent.82
Especially, Article XXIV of the GATT has been the subject of much controversy and disagreement.83 Besides what little clarification the old examination reports, the rare case law, and the Understanding on the Interpretation of Article XXIV of the GATT (hereinafter “RTA Understanding”) offer,84 there has not been an authoritative unambiguous interpretation.85 On
the other hand, paragraph 2(c) of the Enabling Clause has not sparked much debate at all. It has remained mostly absent from WTO case law. This silence also allows for uncertainty on the interpretation. If at least one thing is clear, it is that the interpretation of the provisions of WTO law must be conducted in accordance with the customary rules of interpretation of public international law.86 Therefore, the Vienna Convention on the Law of Treaties (VCLT) applies.87 Pursuant to Article 31(1) of the VCLT, treaty terms must be interpreted in its ordinary meaning in light of the context, and object and purpose. Since Article XXIV of the GATT and the Enabling Clause are exceptions to MFN,88 it follows that these provisions must be interpreted in the context of Article I:1 of the GATT.
The focus in this chapter lies on the interpretation, especially of Article XXIV. Paragraph 3.1 explores the internal requirement for CUs and FTAs to eliminate duties on substantially all trade. Paragraph 3.2 concerns the external requirement for CUs to establish substantially the same external trade policy. Paragraph 3.3 discusses the economic test comparing trade barriers to third countries before and after formation of the RTA. Finally, paragraph 3.4 addresses the role of the Enabling Clause in establishing the GFTN.
82 BM Hoekman and PC Mavroidis, ‘WTO “`a la carte” or “menu du jour”? Assessing the Case for More
Plurilateral Agreements’ (2015) 26 The European Journal of International Law 319, 324.
83 NJS Lockhart and AD Mitchell, ‘Regional Trade Agreements under GATT 1994: An Exception and Its
Limits’ in AD Mitchell (ed.), Challenges and Prospects for the WTO (Cameron May, London 2005).
84 Adopted in the Uruguay Round as part of the GATT 1994.
85 AO Sykes, ‘The Law, Economics and Politics of Preferential Trading Arrangements: an Introduction’ (2010)
46 Stanford Journal of International Law 171, 172
86 Understanding on Rules and Procedures Governing the Settlement of Disputes, Annex 2 Agreement
Establishing the World Trade Agreement (15 April 1994) 1869 U.N.T.S. 40 (hereinafter “DSU”), art 3(2).
87 Appellate Body Report, United States – Standards for Reformulated and Conventional Gasoline, (adopted 20
May 1996) WT/DS2/AB/R, 23; Vienna Convention on the Law of Treaties (23 May 1969) 1155 U.NT.S. 331 (hereinafter “VCLT”).
88 Appellate Body Report, EC – Conditions for the Granting of Tariff Preferences to Developing Countries
(adopted 20 April 2004) WT/DS246/AB/R [90]; Appellate Body Report, Turkey – Restrictions on Imports of Textile and Clothing Products (adopted 19 November 1999) WT/DS34/AB/R (hereinafter “ABR, Turkey – Textiles”), footnote 13.
3.1 Internal Requirement
For CUs and FTAs, Article XXIV:8(a)(i) respectively 8(b) of the GATT provide for an internal requirement, consisting of 3 elements: (i) elimination; (ii) of duties and other restrictive regulations of commerce; (iii) on substantially all the trade between the constituent territories. The first two elements are relatively straightforward. Article I:1 of the GATT speaks of ‘customs duties and charges’, so it is clear that RTAs can include customs duties, thus tariffs.89 The meaning of ‘other restrictive regulations of commerce’ is much more
ambiguous, but can be neglected in this paper, which only focuses on tariffs.90
‘Eliminated’ can be read in its ordinary meaning ‘completely remove or get rid of’.91 It is also
informed by the term ‘reduction’ used in paragraph 2(c) of the Enabling Clause. Accordingly, a nonzero tariff duty would not be consistent with the internal requirement of Article XXIV. In that case, the RTA does not establish free trade and it is not one of the pieces that make up the GFTN. Still, such RTAs may be justified by the Enabling Clause. This will be further discussed in paragraph 3.4.
It may seem contradictory that Article XXIV only authorizes derogation from MFN if it goes all the way, but this can be understood from the history of CUs as step-up to single nations as discussed in chapter 2.92 Additionally, the high political sacrifice prevents WTO members from easily abusing the exception to MFN.93 The minimum threshold of product coverage inherent in ‘substantially all the trade’ also deters WTO members from freely derogating from MFN.94 There is no consensus on the precise meaning of ‘substantially all the trade’, but it is agreed that it includes quantitative and qualitative aspects.95 The qualitative aspect concerns the issue whether the RTA must liberalize trade in every major sector of the economies of constituent parties.96 This will be further discussed in paragraph 3.1.1. The quantitative aspect can be illustrated with the following description of the FTA between the ASEAN and India (AIFTA): the AIFTA establishes on average tariff-free trade on 73.0 percent of the tariff lines, but the percentage varies per country, from 46.5 percent (Indonesia) to 84.5 percent
89 WTO, ‘Tariffs’ <https://www.wto.org/english/tratop_e/tariffs_e/tariffs_e.htm> accessed 27 October 2017. 90 See for a discussion of the interpretation of ‘other restrictive regulations of commerce’ J Mathis, ‘Regional
Trade Agreements and Domestic Regulation: What Reach for “Other Restrictive Regulations of Commerce”?’ in L Bartels and F Ortino, Regional Trade Agreements and the WTO Legal System (OUP, Oxford 2006).
91 English Oxford Living Dictionaries, ‘eliminate’ <https://en.oxforddictionaries.com/definition/eliminate>
accessed 30 October 2017.
92 Bhagwati 1993 (n 12), 25. 93 Chase 2006 (n 18), 28.
94 Hoekman and Mavroidis 2015 (n 81), 323.
95 Working Party Report, European Free Trade Area – Examination of the Stockholm Convention (adopted 4
June 1960) GATT Document L/1235, BISD 9S/70 [48]-[49]; PR, Turkey –Textiles (n 79) [8.5].
(Cambodia).97 Three issue can be raised based on this description. First, what percentage does
‘substantial’ represent? Second, should the internal requirement be assessed on average concessions in the RTA or per constituent party? Finally, is the basis of the assessment the tariff lines or actual trade flows? These issues are addressed in paragraph 3.1.2.
3.1.1 Qualitative aspect
In the Working Party’s examination of the EFTA, concerns were raised about the EFTA’s exclusion of agricultural products from the FTA. The Working Party considered that this exclusion made the RTA inconsistent with Article XXIV:8(b).98 The EFTA countries contended that the flexibility of ‘substantially all trade’ instead of ‘all trade’ was included so countries would be able to protect sensitive products. Article XXIV does not require the RTA to concern ‘trade in substantially all products’. At the same time, they argued that Article XXIV should not be interpreted in a way that required all constituent parties to eliminate tariffs on a certain product. The individual constituent parties should be allowed to exclude different products, even if all excluded products fall within the same sector.99 In the end, the Working Party was unable to conclude how the internal requirement should be interpreted. Nowadays, the preamble of the RTA Understanding states that FTAs and CUs contribute to world trade, and that this contribution is diminished if any major sector of trade is excluded. This seems to point towards the view that Article XXIV does not allow the exclusion of major sectors. In that case, Lockhart and Mitchell (2005) point out that it is necessary to formulate criteria to determine what constitutes a major economic sector. Most likely, these criteria will contain a quantitative aspect, such as a minimum value or volume of actual trade.100
From the perspective of the GFTN, a strict qualitative internal requirement is desirable. After all, only if all constituent parties eliminate tariffs on a given product, will that product be traded freely. Whether the excluded products are concentrated in one sector or spread out over all sectors, does not matter to the GFTN. For the GFTN it is mainly important whether RTAs involve symmetric concessions as assumed in paragraph 2.2. Furthermore, the discussion of the qualitative aspect makes clear that the GFTN can be examined per product. The assumption in paragraph 2.2 that RTAs cover all products allowed this to be neglected, while in fact, there can be as many GFTNs as there are tariff lines.
97 KN Harilal, ‘Regional Route to Multilateralism: Proliferation of PTAs among Developing Countries and WTO
Negotiations’ (2012) 3 Millennial Asia 1, table 1.
98 L/1235 (n 95) [48]. 99 L/1235 (n 95) [51], [54].
3.1.2 Quantitative aspect
The first interpretative issue on the quantitative aspect of ‘substantially all the trade’ that arises is whether 73.0 percent is sufficient. Unlike ‘eliminated’, ‘substantially’ in its ordinary meaning is not associated with a clear number. This does not change with the Appellate Body’s statement in Turkey – Textiles that ‘substantially’ is ‘not the same as all’ but ‘considerably more than merely some’.101 The European Economic Community (EEC) had
suggested that ‘substantially’ should mean at least 80 percent of total trade. The Working Group on the EEC-OCT Association, now EU-OCT Association, rejected this view and argued that ‘substantially’ should be interpreted case by case.102 It was proposed in the CRTA to further clarify the internal requirement as 95 percent of the tariff lines maintained by the Harmonized Commodity Description and Coding System (HS), but no agreement was reached.103
The importance of this issue to the GFTN is best explained with the scenario of a world with 4 actors (A, B, C, D) and 10 products (numbered 1-10). Accordingly, the GFTN would consist of 6 unique symmetric bilateral ties. Table 1 in the Annex shows that in the proposed scenario the GFTN is achieved for 1 product if on average each bilateral tie concerns 8.5 out of 10 products. Therefore, substantial would mean 85 percent. But if substantial meant 80 percent, no GFTN is established. If substantial meant 90 percent, 4 products would be traded freely.
The second interpretative issue concerns the object of the assessment, namely whether the average preferences granted by the RTA or the preferences granted per constituent territory, was decisive. Assume for now that ‘substantially all the trade’ means 70 percent of tariff lines. Based on individual concessions, the AIFTA would not be fully consistent, because Indonesia only eliminates 46.5 percent of tariff lines and Viet Nam 69.2 percent.104 Based on average concessions, the AIFTA would comply as the overall tariff elimination covered 73.0 percent of tariff lines. Another dimension is added if a contracting party to the RTA is a RTA itself, which is the case of the AIFTA. It is possible to say the ASEAN concedes 73.9 percent based on the individual concessions of the ASEAN countries.105 This approach must be taken for the EU based on Article XXIV:1 of the GATT, which determines that the GATT applies to customs territories that have accepted or apply the GATT. As explained in paragraph 2.2.1 the
101 ABR, Turkey – Textiles (n 88) [48].
102 Working Party Report, The Overseas Association, (adopted 29 November 1957) GATT Document L/778,
BISD 6S/70 [30], [34].
103 CRTA, Note on the Meetings of 23-24 September 1998 (16 October 1998) WT/REG/M/19 [12]-[22]. 104 Harilal 2012 (n 97), table 1.
EU can be considered as 1 actor, or 1 customs territory. The EU as RTA is unique in that sense. Other RTAs do not justify for such treatment due to the lesser degree of integration on external trade matters. For example, Thailand has concluded a FTA with Chile while being party to the ASEAN. Therefore, if the requirement of ‘substantially all the trade’ applies to individual concessions, the ASEAN countries must be assessed separately.
For the achievement of the GFTN it does not matter which approach is taken. Once again, imagine a world with 10 products and 4 actors, who now want to conclude 1 multilateral RTA. To reach an average of 80 percent liberalization of tariff lines, the total number of concession must be 32 and each country can agree to at most 10 concessions. Regardless of how these concessions are distributed over the 4 actors, the GFTN is achieved for at least 2 products. Table 2 in the Annex shows what happens if the internal requirement applies per country. Again, mutual free trade covers at least 2 products.
The third interpretative issue also concerns the object of the assessment. The preferential trade liberalization of the AIFTA has been expressed in terms of tariff lines, but the examination of the EEC-OCT Association actually looked at real trade flows.106 One of the reasons why aforementioned proposal in the CRTA to quantify the internal requirement was rejected, was mainly due to its focus on tariff lines. First of all, actual trade between constituent parties may be concentrated in a few tariff lines, which also stresses the importance of the qualitative aspect. Second, individual countries may use different tariff schedules that deviate from the HS. Third, the HS tariff lines may not be the appropriate standard, because a quarter of the tariff lines dealt with agricultural products.107 However, it should be pointed out that an
assessment of tariff lines is much easier than an assessment of trade flows. In case of trade flows, the question rises how they should be calculated. Should the assessment concern actual trade flows or expected trade flows in the situation of free trade? Should trade flows be expressed in terms of volume or value? What is the reference period over which trade flows should be calculated: perhaps 1 or 5 years before or after the RTA was formed?
The EEC-OCT Association raised another red flag, namely whether the assessment concerned trade volumes between each OCT and each EEC member state, or between the OCT and EEC as whole.108 With regard to the EEC, now EU, it was already explained that it constitutes 1 actor. Pursuant to Article XXIV:1 of the GATT, the internal requirement would apply to the
106 L/778 (n 102), Annex IV [35].
107 WT/REG/M/19 (n 103) [15]-[16], [18]. 108 L/778 (n 102), Annex IV [17].
EU as whole, but this is not the case for the OCT. Thus, if the second and third issue are considered in conjunction, the solution is clear.
The question whether imports from A to B and imports from B to A form one or separate trade flows is more complicated. Lockhart and Mitchell (2005) suggested that Article XXIV would allow unilateral RTAs if the aggregate approach is taken. Imagine that substantial is 95 percent and A imports $95 million worth of products from B, while B imports $5 million worth of products from A. If A eliminates tariffs on $95 million worth of trade, B does not need to make any concessions yet the RTA would be in compliance with the internal requirement.109 The reasoning is the same for tariff lines instead of trade values. However, the proposed possibility is prevented by the preamble of the Transparency Mechanism on RTA, which stipulates that RTAs have a ‘mutually preferential nature’. Meaning both A and B must grant each other some kind of preferential tariff treatment, so B needs to eliminate tariffs on at least 1 product even if it is not actually imported.
It is logical that such valueless concessions are undesired. The sole focus on tariff lines would facilitate such empty RTA concessions. Assume for example that country A only imports 1 out of 10 existing product from B. If A eliminates tariffs on products 2 to 10, it has liberalized 90 percent of trade, but it would not have any actual effect. An eventual GFTN would similarly be meaningless.
Nonetheless, B’s concession may also prompt producers in A to switch to the freely tradable product. In that case the RTA may be consistent if the object of assessment is actual trade after formation. However, this would mean the RTA’s compliance with WTO law is assessed in retrospect on its effect. It would be impossible to know before the formation whether a RTA is in compliance. Although enforcement does not really pose a threat to RTAs in the current WTO system, such an interpretation would increase legal uncertainty.
A focus on actual trade flows is supported by the fact that ‘substantially all’ in sub-paragraphs 8(a)(i) and 8(b) does not qualify ‘duties and other restrictive regulations of commerce’, but qualifies ‘the trade between the constituent territories’. Therefore, the emphasis lies on the amount of trade rather than the number of duties or tariff lines. Further support can be found in Turkey’s submission in Turkey - Textiles that 40 percent of its exports to the EU concerns textile and clothing products would also support this view.110 Somewhat similar is the consideration that the automotive sector makes up 29 percent of intra-MERCOSUR trade.111
109 Lockhart and Mitchell 2005 (n 83), 233-234. 110 See PR, Turkey – Textiles (n 79) [17] 111 PR, Brazil - Tires (n 79) [4.378].
Further support for an assessment on actual trade, is provided by the Transparency Mechanism on RTAs. For the assessment it is necessary to know which products are traded, as well as the volume and value of the trade. The Transparency Mechanism for RTAs already requires such information to be provided for the most recent 3 years preceding the notification.112 It follows that 3 years prior to notification is deemed as an appropriate reference period.
Nevertheless, a sole focus on actual trade flows would not be sustainable from the perspective of the GFTN. In the same scenario that A only imports product 1, the RTA would be consistent if it liberalized 10 percent of tariff lines. If A’s producers of product 1 realizes economies of scale, a trade creating RTA would switch demand in A from product 1 imported from B to domestically produced product 1. Producers in country B would have to switch exports to A to other products like product 2. In the long run, product 1 will only make up a small portion of actual trade. A new RTA covering product 2 traded from B to A would be necessary.
If no new RTA is concluded, A and B may consider an amendment of the old RTA. Whether this is allowed under Article XXIV can be doubted. Paragraphs 8(a) and 8(b) of Article XXIV should be read together with the chapeau of paragraph 5. While paragraphs 8(a) and 8(b) provide for the definition of CUs and FTAs, it is actually paragraph 5 which authorizes the formation of CUs and FTAs. The Appellate Body in Turkey – Textiles had interpreted the chapeau of paragraph 5 to mean that measures, otherwise inconsistent with the GATT, are only exempt if they are introduced upon the formation.113 Measures added to the RTA after
formation would not fall within the scope of the exception.114 If the amendment concerns an additional tariff line on which tariffs are eliminated, this interpretation does not sit well with the purpose of the GFTN. The preamble of the RTA Understanding clarifies that the purpose of Article XXIV is to eliminate all tariffs on trade between constituent parties. If the amendment is a move towards the goal of the exception, it does not make sense to include that move from the exception.
Conversely, the question can be raised what happens if RTAs no longer meet the internal requirement. For example, trade flows may change after formation and the RTA no longer covers a substantial portion of actual trade. Would that mean the RTA is no longer a justified deviation from MFN? No. Article XXIV:5 stipulates the GATT does not prevent the
112 TM RTA (n 75), [7(a)] in conjunction with Annex [2(e)]. 113 ABR. Turkey – Textiles (n 88) [46].
‘formation’ of a FTA or CU. As long as the text of the RTA does not change, it would seem that the effect, application, and enforcement fall outside the scope of Article XXIV:5.
3.2 External requirement
Article XXIV:8(a)(ii) provides for an external requirement for CUs that requires constituent parties to apply ‘substantially the same duties’ to the trade of third countries. Like the internal requirement, its origins lie in the historical perspective of CUs having the nature of single nations.115 The Appellate Body in Turkey – Textiles explained the RTA’s common external
trade regime need not be the same, but should show ‘a higher degree of “sameness”’ than ‘comparable trade regulations with similar effects’.116 As with ‘substantially all the trade’,
‘substantially the same duties’ comes with qualitative and quantitative aspect.117 Hence, the
interpretative issues of the internal requirement apply mutatis mutandis to the external requirement.118 The external requirement has the added complication that it concerns tariffs on trade with third countries, and need not be zero. Accordingly, the flexibility in the external requirement can be utilized by CU member states in three ways: in terms of tariff rates, traded products, or trading partners.
Imagine the world exists of 12 actors, 10 products and ‘substantially’ means 90 percent. Suppose a CU is concluded between actors A and B, meaning there are 10 third countries. This CU sets the common external tariff rate at 5 percent. However, CU members could apply tariff rates between 4.5 percent and 5.5 percent, since this is within the maximum deviation of 10 percent. Additionally, CU members can apply the common external tariff of 5 percent on trade with 9 of the third countries, but only for 9 products. The 10th product is subject to individually set tariff rates. With regard to the 10th country, the CU members may apply different tariff rates.
The different rates on trade between the CU members and the 10th third country may be pursuant to a RTA between that third country and CU member A. After all, Article XXIV does not determine that WTO members can only be party to one RTA. It only stipulates requirements for the formation of individual RTAs, but is silent on the interaction between multiple RTAs. The EU aside, most CUs do not take away the autonomy of constituent parties over external trade matters. In fact, many CU member states have also formed FTAs.
115 Bhagwati 1993 (n 12), 25-26.
116 ABR, Turkey – Textiles (n 88) [49]-[50]. 117 PR, Turkey – Textiles (n 79) [9.148]. 118 Lockhart and Mitchell 2005 (n 83), 248.
