Journal of International Economic Law Advance Access originally published online on April 28, 2007
Journal of International Economic Law 2007 10(2):317-334; doi:10.1093/jiel/jgm008
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© Oxford University Press 2007, all rights reserved
Independence Under Fire: Extra-legal Pressures and Coalition Building in WTO Dispute Settlement
*John Maton is a trainee solicitor with Trowers and Hamlins, London, and an MSc graduate of the Department of Political Science/School of Public Policy, University College London. Carolyn Maton is a junior doctor, and an MA and MB/Bchir graduate of Clare College, University of Cambridge.
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONI. THE LITERATUREII. METHODOLOGY - THE...III. RESULTSIV. ANALYSISV. CONCLUSIONS AND... |
|---|
This study uses detailed quantitative analyses of the complete history of rulings made by the Panels and the Appellate Body of the WTO Dispute Settlement Mechanism during its first ten years of operation, to assess the robustness of theories regarding the decision-making of these institutions. Regression analyses are used to test for correlation between the success of Complainants in dispute settlement and a variety of factors, representing the hypothesized capacity of states to influence Panels and the Appellate Body through dimensions of political and economic power, the impact of the relative practical capacities of states in dispute settlement proceedings, and the formation of coalitions of states in support of a particular Complainant or Respondent. The results of the regression analyses – that few significant correlations exist between the independent variables and the dispute outcomes – provide evidence that the judicial institutions of the Dispute Settlement Mechanism are independent from Member State influence. The one reliable correlation found in these analyses demonstrates that a Complainant state does have substantial advantage in Panel proceedings if it has previously participated in more disputes than the Respondent.
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONI. THE LITERATUREII. METHODOLOGY - THE...III. RESULTSIV. ANALYSISV. CONCLUSIONS AND... |
|---|
Founded on 1 January 1995, the World Trade Organization (WTO) is one of the most important and controversial organizations in global economic policy. Although the familiar depictions of it as some all-powerful nemesis of the global public good can be dismissed as populist hyperbole,1 a number of serious objections to the rules and practices of the Organization do exist. In particular, it is argued that the Organization is institutionally dominated by its more powerful Member States, and that its apparent equal treatment of all members masks inherent bias towards the industrialized world.2
Nowhere in the WTO regime is institutional independence from Member States more important than in the Dispute Settlement Mechanism (DSM), the quasi-judicial system through which alleged violations of WTO agreements are adjudicated. Unprecedented steps were taken by the Parties to the Uruguay Round Agreements, establishing the Organization, in attempting to create an international arbitration system free from the influence of diplomatic power relations, under which the force of legal argument alone could prevail. This study examines the results so far.
The central focus here is the posited capacity of states, as parties to formal dispute settlement proceedings, to influence the outcome of these proceedings other than by legal argument, and thereby to attack the trade policy measures of other states, or to defend their own. It assesses both the individual power of states in this regard, and the influence of third party states intervening in support of others. It is hoped that the outcome will make a novel contribution to the academic study of the DSM, and develop understanding of the system at the practical level, as well as informing the public debate about state sovereignty, international law and the concept of ‘world government’.
The approach taken in this study adopts techniques from both legal studies and political science scholarship to explore the workings of the Organization's legal machinery, and to undertake a sympathetic but sceptical analysis of political science criticisms of the DSM. This study is neither an attempt to prove nor disprove any claims made by critics of the WTO, but rather an attempt to carry out one of the first complete analyses of the performance of the DSM, and to assess what evidence there may be that states can influence its operation and decisions. The results permit a rigorous and comprehensive study of patterns in dispute outcomes, as well as qualitative observations about the DSM on the basis of the data available.
In attempting a quantitative analysis of the complete history of WTO dispute settlement, this study examines the potential for a more comprehensive survey of the outcomes of DSM litigation than has yet been made. It is hoped that the results will illuminate the patterns of dispute outcomes in the past, and shed light on an area of international relations and public policy which is all too often misunderstood.
This article proceeds as follows. Section I reviews the literature in this area of study. Section II describes the research methodology used. Section III firstly analyses some descriptive statistics on the system and dispute outcomes, commenting on the incidence of Complainant wins and the success rates of the United States and EC; and secondly discusses the results of the regression analyses which are the focus of this article. Section IV interprets the results further, and Section V concludes.
|
I. THE LITERATURE |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONI. THE LITERATUREII. METHODOLOGY - THE...III. RESULTSIV. ANALYSISV. CONCLUSIONS AND... |
|---|
The mood expressed in the plentiful literature on the multilateral trading system is generally, although not unanimously, approving of the move from diplomacy to legalism made in the foundation of the WTO.3 In the decade since its creation, much praise has been given both to the new DSM and those responsible for engineering it,4 for the increased institutional independence and efficiency that it has brought about.
The two developments which are most often selected for particular praise are the principle of negative consensus, and the introduction of a standing Appellate Body as the final arbiter on WTO disputes, both of which have had the effect of removing practical authority over the dispute settlement process further from the Member States, and are therefore said to have enhanced the level of delegation in the system, and in consequence, its independence from the parties.5 Indeed, there is some debate over whether the organs of the DSM, in particular the Appellate Body, are too insulated from Member State supervision in what is, after all, a diplomatic forum.6 However, some have characterized the DSM's independence as only ‘moderate’, due to the influence that Member States retain over the system through, for example, the selection of Panellists.7
Criticisms that the DSM is in some way biased in favour of rich countries, multinational corporations or other dominant actors, common in mainstream debate, are not often found in serious legal-political literature, but some such arguments have been made.
Garrett and McCall-Smith, focusing on the Appellate Body, claim that the pressures created by the institutional structure of the WTO judicial system favour the more powerful Member States. They examine perceived ‘conciliation’ by the Appellate Body, which they see as the practice of allowing political considerations to take precedence over legal reasoning when choosing whether to rule against politically powerful Respondents. They argue that ‘in the short term ... decision making will be strategic and often political’, and provide evidence to support this, stating that they ‘expect the Appellate Body to be reluctant to make strong and unequivocal adverse rulings against powerful WTO members on issues of considerable domestic political salience [for those members]’.8 However, their analysis is based only upon the qualitative examination of particular decisions, rather than an overview of the system as a whole.
Steinberg criticizes the DSM in similar terms. He argues that judicial decision-making in the WTO faces ‘a hard political constraint’ in the form of the interests of powerful Members, particularly the United States and EU,9 and that these actors are able to exert significant political influence over the rulings of the Appellate Body. Writing on the history of ‘judicial lawmaking’ by the Appellate Body, he claims that the more powerful actors within the WTO system are able to influence the DSM to their own advantage.
Steinberg cites certain institutional characteristics in particular:
Appellate Body members are selected through a process in which the powerful members may veto candidates whom they assess as likely to engage in inappropriate or undesired lawmaking; the Appellate Body acts in the shadow of threats to rewrite DSU rules that would weaken it and of possible defiance of its decisions by powerful members; and the Appellate Body receives – and has established means of obtaining - information on the preferences of powerful members, helping it to avoid political pitfalls10
However, as with Garrett and McCall-Smith, Steinberg does not base his argument on analysis of the overall history of dispute settlement.
This is not to say that either of the above criticisms necessarily lack validity. It is clear that the elements of the system cited by Steinberg, and the events surrounding particular disputes can give rise to the suspicion that political influence may indeed be a factor in the decision-making of the DSM institutions. It is criticisms of this nature which this study aims to assess.
A comprehensive study of the patterns of DSM decisions has yet to be conducted to assess the factual pattern of outcomes of cases under the new system; to date, statistical analyses of the WTO have generally been limited to uncontroversial matters of fact.11 Those writers who have attempted to understand and explain the outcomes of DSM decision-making have purported to prove a number of theories, but only a few are grounded in robust statistical evidence.
Of the studies which have been published of this kind, the work of Busch and Reinhardt is the most similar to that presented here. They construct a model to assess the influence of third parties on proceedings in the DSM. Their work is detailed and highly instructive on the potential influence of third parties in the DSM process. They find that third party participation reduces the likelihood of early settlement in a given dispute, and, importantly for the direction of this study, that ‘the conventional wisdom that third parties influence rulings is correct, but only because their participation dramatically lowers the odds of a negotiated solution’.12 To the authors’ knowledge, this is the closest any previous study has come to an analysis of the factors affecting the legal outcome of WTO disputes.
This study takes a different approach to that taken by Busch and Reinhardt, in that it combines an assessment of third party influence with the central element of the study, an analysis of what effect the power and capacities of states may have on the substantive legal rulings of Panels and the Appellate Body.
Other studies on related topics include work by Bown, who presents evidence that institutional factors may impact upon the decision by a state as to whether or not to participate in dispute settlement.13 Participation is less likely, he finds, by smaller countries who are likely to face retaliation, or who are particularly reliant on the potential Respondent in a particular dispute. Guzman and Simmons14 use quantitative methods to assess the patterns of selection of Respondents by states in DSM litigation. They find that less politically powerful states are likely to choose to litigate against the more powerful states, as the expected returns from such proceedings (economically speaking) are higher. The same authors have also analysed the factors affecting the decision by states on whether or not to settle prior to full Panel proceedings.15
Few studies of the factors affecting judicial decisions have been conducted at the international level at all, although a plentiful literature exists on the decision-making of courts and quasi-judicial bodies in domestic legal systems, largely in the United States.16 Consideration of these studies is outside the focus of this literature review, but the formats of these analyses have greatly influenced this study.
Those writers who have previously analysed the DSM sometimes refer to evidence that domestic civil litigation will, on average, result in approximately 50% success for claimants.17 Comparing this to the history of GATT and WTO dispute settlement, Complainants have had far greater success than the theory predicts.18 Reasons posited for this include a pro-free trade bias in DSM rulings,19 and a lack of willingness on the part of Complainants to progress with claims of which they are not fully certain, due to the political costs involved.20 Quantitative techniques have also been used to examine the incidence of solutions being reached at the consultation stage of GATT proceedings, with the finding that this is more likely to occur when both parties are democracies.21
However, the central question remains – in what ways, if at all, can Member States influence the outcome of contentious dispute settlement proceedings? Some qualitative studies have highlighted ways in which the organs of the DSM may be pressurized by parties to disputes, or by extraneous factors. These include the possibility that Panels and/or the Appellate Body may favour more powerful states, so as not to alienate the most significant actors in the WTO,22 or that pressure may be exerted on the formally independent judicial branch of the Organization by the Member States, through the ‘political’ negotiating branch.23
One of the few quantitative studies of this kind to date, which analyses GATT dispute rulings, finds a (counterintuitive) trend towards decisions in favour of states with lower GDP values, controlling for other variables.24 However, the confidence intervals in the study were acknowledged to be too wide to permit any firm conclusions.
Some writers maintain that the organs of the DSM are not vulnerable to extra-legal influence;25 however, to the author's knowledge, such assertions have yet to be backed by either statistical evidence or a comprehensive survey of all dispute rulings. In addition, analyses of DSM and other international jurisprudence often focus on examining the institutional characteristics and preferences of judicial bodies, rather than on tendencies in their substantive decision-making. Without such evidence, it is difficult to either support or rebut criticism from those who do not subscribe to the view of the DSM as an impartial arbiter of WTO law.
|
II. METHODOLOGY – THE RESEARCH MODEL |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONI. THE LITERATUREII. METHODOLOGY - THE...III. RESULTSIV. ANALYSISV. CONCLUSIONS AND... |
|---|
The first step in constructing the model for this study was the selection of the appropriate unit of analysis. In order to achieve comprehensive results, the study was conducted using the complete history of substantive Panel and Appellate Body decisions on disputes under the WTO regime, with the following exceptions: the list was limited to disputes in which a substantive decision had been reached by 31 December 2004, due to the lack of sufficient economic data after this point; the analysis excludes reports which simply confirm a mutually agreed solution between the parties; and, in order to preserve the clarity of the model, rulings following challenges under Article 21.5 DSU26 are also omitted.27
One important observation regarding the unit of analysis is that the disputes as categorized by the WTO do not correspond precisely to the number and range of trade policies which have been litigated through the DSM.28 Some disputes involve claims brought by multiple Complainants,29 whereas others may involve the same Respondent and the same issues, but be divided into separate disputes.30 Conceptual consistency cannot therefore be maintained by analysing all disputes, classified by their WTO classification numbers, as equal units, due to the potentially influential effect of the intervention of multiple participants.
In this study, the unit of analysis retains the WTO's numerical classification system, but divides disputes involving multiple Complainants into several bilateral disputes, one for each Complainant,31 in each case treating the other Complainants as third parties, to be included at the secondary level of analysis. This method enables a detailed examination of bilateral legal relations, and, with the addition of third party analysis (described further), does not ignore the impact of multiple Complainants.
A. The independent variables
The central issue for this study is whether, and in what ways, it is possible for Complainant states to influence the outcome of disputes through extra-legal means. The independent (input) variables used to test this hypothesis were chosen to reflect the range of ways in which the parties to a dispute could conceivably influence the outcome of Panel or Appellate Body proceedings.
It is suggested that this hypothetical influence may be in the form of either deliberate action taken by states, or strategic choices made by Panels or the Appellate Body, and that it can be classified into three categories – Political Power,32 Economic Power33 and Practical Capacity.34 From these three categories, a set of four independent variables (two continuous variables and two dichotomous) was derived, forming the first level of the model.35
The continuous variables were differentials between the Complainant and Respondent in (i) the ‘Economic Power Index’, developed as described further, and (ii) previous participation in WTO disputes, up to the date of the initial request for consultations.
Initially, differentials in the GDP and trade volumes of the respective states were used as separate independent variables, as both could hypothetically have an impact on the decision making of Panels or the Appellate Body. It was found in the initial stages of the regression analyses that these two variables exhibited extreme multicollinearity – their patterns of variance were found to be too similar for any observed variation in outcomes to be attributed to any one input variable. Therefore, an index variable was constructed to take account of these factors.
This index was achieved by calculating the sum of the two factors for each state, finding the mean average of this figure over the time period of the study,36 and dividing each mean by the lowest mean figure calculated from sufficient reliable data.37 The differential between this ‘Power Index’ figure for the Complainant and Respondent in each dispute was then applied as an independent variable.
It was hypothesized that if the potential Political, Economic and Practical channels of extra-legal influence do in fact have an impact on DSM decision-making, the greater the (positive) difference in Power Index value or previous use of the system between the Complainant and Respondent, the more likely the decision in a particular dispute is to favour the Complainant.
The second continuous variable, the differential in previous use of the DSM, was derived simply by calculating, for each dispute, the number of disputes before it in which each party had participated, and subtracting the Respondent's figure from that of the Complainant. In each case, the date of the initial request for consultations was used to determine chronological order.
The dichotomous variables used were based upon the identity of the Complainant and Respondent in each dispute. On the hypothesis that the EC and the US are more likely than other states to be successful, due to their disproportionate political leverage in the context of international trade, the dichotomous variables were: (i) was the dispute a claim by the EC or the United States against a third-country Respondent which was not one of these parties; and (ii) was it a claim against one of these parties by a third-country Complainant? It was further hypothesized that, if claims such as those reviewed earlier, about systemic bias and extra-legal influence are true, then the EC or United States may be in a stronger position in DSM litigation.
B. The dependent variable
In coding the dependent (outcome) variable in this study, the results of disputes, decisions of the Panel or Appellate Body were coded according to whether the policies under challenge were found to be consistent or inconsistent with the Respondent's obligations under the covered agreements. Purely procedural decisions with no direct impact on policy were not included. The scheme by which results were coded is of fundamental importance to the structure of this study, and therefore merits explanation in some detail.
Due to the complexity of many disputes litigated through the DSM, the coding scheme for results was based around individual arguments. For these purposes, an ‘argument’ is a claim by a WTO Member State that a trade policy measure of another state (not an action during dispute settlement proceedings) is inconsistent with that state's obligations under the covered agreements. Claims regarding legal exceptions which may legitimize inconsistencies alleged in other arguments, or claims made in the alternative to others, are not considered separately.
This scheme for categorising arguments was then applied to the conclusions of the Panel/Appellate Body, as described in the relevant published report,38 to assess the number of arguments in which each party is successful. Each argument under which the Respondent is found in violation of its obligations was awarded to the Complainant. Arguments in which a claim fails, or which the Panel or Appellate Body refuses to consider for procedural reasons are awarded to the Respondent.39 Elements of the ruling which do not directly relate to the consistency of a trade policy measure with WTO agreements are ignored for the purposes of this scheme.
The primary form of analysis used was logistic (logit) regression, using a binary dependent variable.40 Results were coded as a ‘win’ or ‘loss’ for the Complainant, depending upon whether or not in the Complainant won as many or more arguments as the Respondent.41
Previous quantitative analyses have coded results in a variety of ways, including: dichotomously, according to whether or not violations are found;42 in three categories, as rulings favouring the Complainant or Respondent, and mixed rulings;43 or according to the eventual policy outcome of the dispute as a whole, with the formal ruling being only one element of this.44 The coding scheme used here focuses in more detail on individual legal arguments than previous analyses, and provides a more sophisticated analysis of the formal results of DSM proceedings than has been carried out to date.
C. Incorporating third parties
The second level of analysis in this study examines the impact of third party ‘coalitions’45 on DSM proceedings. This was assessed both as a new variable in itself, specifically the differential between the raw numbers of third parties supporting each participant, and also by weighting this figure to take into account the totals of the Power Index and previous use figures relevant to the states in coalitions.
In order to assess the impact of coalition building on dispute settlement proceedings, it was necessary to create a new coding scheme for third party positions, and this was done on the basis of their submissions and answers to questions, as presented in published reports. Third party submissions were coded as favouring the Complainant if they argued that the policies under consideration are inconsistent with WTO agreements; if they presented legal arguments which, if accepted, would render the measures inconsistent; or if they claimed that a ruling favouring the Respondent would have solely negative consequences. Submissions favouring the Respondent were those which argued for the opposite conclusions. Submissions which presented some arguments for either side, or which display none of the above characteristics, were coded as neutral.
This coding scheme, while admittedly not capturing all the nuances of the legal positions involved, provides for a clear and reliable analysis of third party positions regarding the policies under challenge in each dispute.
|
III. RESULTS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONI. THE LITERATUREII. METHODOLOGY - THE...III. RESULTSIV. ANALYSISV. CONCLUSIONS AND... |
|---|
A. Descriptive statistics
The high success rate for Complainants in DSM rulings continues; Complainants ‘win’ (on binary coding) over 80% of all disputes (193 of 240 rulings), combining Panel and Appellate Body data; 81.9% of Panel rulings and 78.4% of rulings by the Appellate Body have been in favour of the Complainant. In fact, applying techniques from previous analyses46 to the data collected for this study, the Complainant success rate has risen between June 2002 and December 2004 – from 88% (52 of 59 rulings)47 for Panels and the Appellate Body combined, to over 92% (222 of 240 rulings). Reasons for this high percentage of success have been discussed elsewhere in the literature, however, and are not the focus of this study.
Regarding the posited influence of the more powerful Member States on DSM rulings, it is also interesting to note the rates of success of the major trading powers in dispute settlement. The United States and the EC, the only two parties to have been involved in enough decided disputes to be analysed, provide interesting results (Table 1).
|
The EC enjoys above average success at all levels – as a Complainant and Respondent, before both the Panel and the Appellate Body. The United States enjoys the same elevated success rate before Panels, but is less likely than average to be successful on appeal. Is this indicative of a pattern of bias in Appellate Body rulings? The regressions in this study, discussed further, suggest not, and the data for both states is surely insufficient for the drawing of any conclusions. However, these anecdotes provide clues for trends which may or may not be borne out in future.
B. Regression analysis: logit
For both the Panel and Appellate Body data sets, logit regressions were conducted, with OLS regression used to support the conclusions drawn.48 This combination of techniques permits the analysis of dispute rulings at both the level of overall outcomes, and that of individual arguments, rather than relying solely on oversimplifications at the generalized level. The logit regression results are reported here, as representative of the findings as a whole.
The independent variables of Difference in Power Index Score, Difference in Previous Use of the System, Difference in Third Party numbers and EC/US against Third-Country were each regressed against the binary dependent variable using the logit method. (As discussed, OLS was used as a secondary method.) These regressions were carried out separately, on a bivariate49 basis, due to the problems of multicollinearity as outlined earlier.
Among the logit regressions conducted, the low Cox & Snell values (measuring the degree of the variation in the dependent variable explained by the independent variable in each case), and high significance values demonstrate that no robust conclusions can be drawn from this analysis.
For the Panel data (Table 2), the Cox & Snell scores showed that no variable explains more than approximately 4% of the observed change in the dependent variable (probability of a verdict for the Complainant), with all but one variable, Difference in Total Index Scores, measuring far less than this. The significance values are all far above the 0.05 level,50 and the relevant null hypotheses of no effect cannot be rejected.
|
From this it can be concluded that none of the variables measured have a significant influence on the outcomes of disputes at the Panel level.
In the Appellate Body logit regressions (Table 3), the Cox & Snell values were, with one exception, all below 0.03, meaning that most of the independent variables explain less than 3% of the variation in the logged odds of a win for the Complainant.
|
As for the Panel results, the significance values for all variables are in excess of the 0.05 level at which a result could be considered statistically significant. Therefore we can see that there is no significant relationship between any of these variables and the outcomes achieved in DSM litigation.
C. Regression analysis: OLS
For the most part, the results of the OLS regression analyses support the findings of the logit stage. Brief mention of two differing results from the OLS analysis is warranted here, however.
The two variables which provided positive results from this analysis are Difference in Previous Use and Difference in Third Party Numbers, in relation to Panel rulings. The significance statistics for these variables are 0.006 and 0.007 respectively, meaning that the null hypotheses of no effect can be rejected at the 0.01 level. There seems, therefore, to be a relationship between these factors and the outcome of Panel proceedings.51
In order to see the effects of these factors on the outcome, controlling for each other, a further regression was carried out using both variables together.52
For each unit increase in the Difference in Use (every previous dispute litigated by the Complainant more than the number for the Respondent), a corresponding 0.237 increase in the percentage of arguments won by the Complainant. For every unit increase in the Difference in Number of Third Parties (every one third party state supporting the Complainant more than the total supporting the Respondent), we see, perhaps surprisingly, a 0.836 decrease in dependent variable of percentage of arguments won by the Complainant.
However, this second result can be excluded, as the Significance figure is 0.061. The same figure for Use Difference is exactly 0.05, so there is a 5% probability that the Difference in Use result is due to chance, although we can (just) reject the null hypothesis in this case, at the 0.05 level.
The R2 model fit statistic for the regression using these two variables is 0.079, meaning that less than 8% of the variation in the dependent variable can be explained by these factors. In conclusion, there seems to be a weak relationship between previous use of the system and the number of arguments won at the Panel stage of WTO proceedings. However, it should be borne in mind that this is not consistent with the logit result for the same variable. Therefore although previous use seems to have an impact on the number of arguments in which a Complainant is successful, it does not correlate with the likelihood of a ‘win’ on this analysis.
|
IV. ANALYSIS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONI. THE LITERATUREII. METHODOLOGY - THE...III. RESULTSIV. ANALYSISV. CONCLUSIONS AND... |
|---|
To summarize, the only apparently significant result is that greater experience of the DSM on the part of the Complainant increases the percentage of arguments won by the Complainant in Panel proceedings.
Greater experience in dispute settlement is a question of practical capacity – where states have participated in a greater number of disputes, their trade ministries and personnel will have greater experience of the system, and hence greater skill at dealing with it, both personally and institutionally. It is interesting to note that this trend is absent in Appellate Body results, suggesting that it has less impact there. This result, however, is subject to several qualifications; including the low level of model fit, and the high significance figure of the results.
The low model fit is perhaps to be expected on the grounds that the model necessarily excludes a number of variables due to multicollinearity. The significance figure is a concern in these results, as the probability of the relationship being due to chance is at the outer bounds of acceptability. However, these issues do not prevent us from drawing the conclusion that there is a positive correlation between previous experience and success at the Panel stage.
On the basis of these results, it may be suggested that the model is poorly specified for this analysis, and that some variables should be removed or added in order to improve it. In anticipation of such criticism, it is submitted that the modelled variables are all required in the study, as all represent different aspects of the range of capacities which states possess to attempt to influence the outcome of WTO proceedings.53 It is acknowledged however that much further study, incorporating the inclusion of a wider range of variables than used here, is necessary to achieve a comprehensive understanding of the factors affecting dispute settlement decisions.
|
V. CONCLUSIONS AND OBSERVATIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONI. THE LITERATUREII. METHODOLOGY - THE...III. RESULTSIV. ANALYSISV. CONCLUSIONS AND... |
|---|
The results of the regression analyses in this study suggest that only one of the extra-legal factors considered has a substantial effect on the decisions of WTO Panels or the Appellate Body (as the other putative correlation is not supported by variables which we would expect to follow the same patterns). The presented evidence suggests that Complainant states which have participated in a larger number of dispute settlement proceedings than their opponents are likely to enjoy a higher success rate in terms of their arguments presented to Panels, as a result of this difference.
In making any statement on the basis of this evidence, we must bear in mind the limitations of the model. The independent variables used in this study provide a broad survey of the dimensions of influence which states may have in the legalized diplomatic arena of the DSM. Unfortunately, it is not possible in a study of this limited scale to take account of all potential factors, or even to fully quantify all variables. For instance, the coding of third party submissions is based purely on the content of dispute settlement reports, using a rigid (and thereby replicable) coding method. It takes no account of states which may support one party or the other, but fail to make submissions. Within the institutional structure of the WTO, such states would certainly be able to make their feelings known on certain issues without fulfilling all of the formalities of the DSM, but the quantification of such aspects of influence would be a task beyond the scope of this work.
It is instructive, however, to note the theories which are undermined by this research. Bearing in mind that the regression analyses in this study constitute a quantitative survey of the entire history of decided WTO disputes, there is no reliable evidence to suggest that either body supports richer or more powerful states as against others, or that they defer to larger coalitions of states on any issue.54 At this level of abstraction, it seems possible to state that the organs of the DSM are immune from such pressures, adding weight to the arguments of the Organization's supporters.
In observing patterns in dispute settlement, it is also of interest to examine the third form of coding which was applied to dispute outcomes, but not used in regression analysis, the ordinal coding for complete, substantial or neutral results, described above.55 In Appellate Body proceedings, mixed results are the most common, representing 49% of all results, as opposed to 32.6% for Panel proceedings, in which complete wins for the Complainant (44.9%) are the most common verdict. It is possible to speculate about the reasons for this, but on the basis of the evidence presented here, no conclusions can be drawn. Further research would be required in order to attempt an explanation of this pattern, and such research is beyond the scope of either this paper or the analysis adopted by it.
This study explicitly focuses on the decisions of Panels and the Appellate Body as they specifically relate to the ‘legality’ of Member State trade policy measures. However, theories of DSM decision-making have often addressed the more complex question of whether the DSM institutions, and the Appellate Body in particular, are pressurized by Member States in their procedural rulings regarding their own jurisdiction and working procedures,56 and any impact such pressure may have is not represented in this analysis. Also, this study says little about any potential institutional bias in the structure of the WTO in favour of powerful Member States, or theories which claim that the very premise of a legalized multilateral trade system is grounded in Western legal and political theory, and that developed countries therefore enjoy a dominant position in the system, at an abstracted theoretical level.57
Finally, it is suggested, on a practical level, that if third party submissions really have no impact whatsoever on the outcome of proceedings, states have no need to continue to make them. There are significant costs in taking part in litigation, and these costs weigh especially heavily on the states least able to afford them. Why take on such costs if there is no increased likelihood of a favourable outcome? Further research, with more complex models and data sets, would be required to confirm this conclusion, but this is to overlook the simplest explanation, and indeed to ignore the overall conclusion of this study: that there is no statistical evidence to support claims of bias or extra-legal influence in WTO dispute settlement.
Supplementary materials are available online at http://jiel.oxfordjournals.org/
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONI. THE LITERATUREII. METHODOLOGY - THE...III. RESULTSIV. ANALYSISV. CONCLUSIONS AND... |
|---|
The authors’ thanks are due to Dr David Coen and Dr Fabio Franchino of the Department of Political Science/School of Public Policy, University College London for their invaluable help with the development of the model and theoretical basis of the work; to Dr Deborah Cass of the London School of Economics and Political Science and Prof Catherine Redgwell of University College London for interviews and comments which greatly assisted the authors’ research into the WTO system; to Arun Venkataraman of the USTR, formerly of the Appellate Body Secretariat, WTO, for help with data collection; and to three anonymous reviewers whose helpful comments have informed revisions made to this article. Any errors remain the responsibility of the authors alone. For questions or comments, the authors can be contacted at Email: john.maton{at}hotmail.co.uk.
1 The anti-globalization NGO Public Citizen describes the WTO as a ‘powerful new global commerce agency ... one of the main mechanisms of corporate globalization [sic]’; see Public Citizen Global Trade Watch website. www.citizen.org/trade (visited 8 August 2006). 
2 See, e.g.: Chakravarthi Raghavan, ‘The World Trade Organization and its Dispute Settlement System: Tilting the Balance Against the South’ (2000), at http://www.twnside.org.sg/title/tilting.htm (visited 11 March 2005).
3 See, e.g.: James Cameron and Kevin R. Gray, ‘The Principles of International Law in the WTO Dispute Settlement Body’, 50 International and Comparative Law Quarterly (2001), 248ff. Claus-Dieter Ehlermann, ‘Tensions Between the Dispute Settlement Process and the Diplomatic and Treaty-Making Activities of the WTO’, 1 World Trade Review (2002b), 301–8. Serge Frechette, C. Michael Hathaway, and Victor Do Prado ‘Performance of the System III: Appellate Body; Comments’, 32 International Lawyer (1998), 747–54. Robert Gilpin, Global Political Economy, (Princeton: Princeton University Press 2001). Andrew L. Stoler, ‘The WTO Dispute Settlement Process: did the Negotiators get what they Wanted?’, 3 World Trade Review (2004) 99–118.
4 See, e.g.: Claus-Dieter Ehlermann, ‘Six years on the Bench of the ‘World Trade Court’: Some Personal Experiences as Member of the Appellate Body of the World Trade Organization’, 36 Journal of World Trade (2002a) 605–39. John H. Jackson, ‘Designing and Implementing Effective Dispute Settlement Procedures: WTO Dispute Settlement, Appraisal and Prospects’, in Anne O. Krueger (ed.), The WTO as an International Organization, (Chicago: University of Chicago Press, 1998), Stoler (2004).
5 See, e.g.: Steve Charnovitz, ‘Judicial Independence in the World Trade Organisation', in Laurence Boisson de Chazournes, Cesare P. R. Romano, and Ruth Mackensie (eds), International Organizations and International Dispute Settlement: Trends and Prospects (Ardsley: Transnational Publishers, 2002).
6 See, e.g.: John Greenwald, ‘WTO Dispute Settlement: An Exercise in Trade Law Legislation?’, 6 JIEL, (2003) 113; Donald McRae, ‘Claus-Dieter Ehlermann's Presentation on the Role and Record of Dispute Settlement Panels and the Appellate Body of the WTO – Comments by Professor Donald McRae’, 6 JIEL (2003) 709ff; cf. William J. Davey, ‘Has the WTO Dispute Settlement System Exceeded its Authority? A Consideration of Deference Shown by the System to Member Government Decisions and its Use of Issue-Avoidance Techniques’, 4 JIEL (2001) 79, Ehlermann (2002a).
7 Robert O. Keohane, Andrew Moravcsik and Anne-Marie Slaughter, ‘Legalized Dispute Resolution: Interstate and Transnational’, 54 International Organization (2000) 457–88 at 461.
8 Geoffrey Garrett and James McCall Smith, ‘The Politics of WTO Dispute Settlement’, paper presented to the Annual Meeting of the American Political Science Association, 1999. http://www.yale.edu/leitner/pdf/1999-05.pdf (visited 8 August 2006) at 44.
9 Richard H. Steinberg, ‘Judicial Lawmaking at the WTO: Discursive, Constitutional and Political Constraints’, 98 American Journal of International Law (2004) 247–75, at 275.
11 See e.g.: Jackson (1998), Kara Leitner and Simon Lester, ‘Notes, Comments and Developments – WTO Dispute Settlement 1995–2004: A Statistical Analysis’, 8 JIEL (2005) 231–44.
12 Marc L. Busch and Eric Reinhardt, ‘Three's a Crowd: Third Parties and WTO Dispute Settlement’ 58 World Politics (2006) 446–477 at 475.
13 Chad P. Bown ‘Participation in WTO Dispute Settlement: Complainants, Interested Parties and Free Riders’, 19 World Bank Economic Review (2005) 287–310.
14 Andrew T. Guzman and Beth A. Simmons, ‘Power Plays & Capacity Constraints: The Selection of Defendants in WTO Disputes’, 34 Journal of Legal Studies (2005) 557.
15 Andrew T. Guzman and Beth A. Simmons, ‘To Settle or Empanel? An Empirical Analysis of Litigation and Settlement at the WTO’, 31 Journal of Legal Studies (2002) 205.
16 See, e.g.: Tom L. Austin and Donald Hummer ‘The Effect of Legal and Extra-Legal Variables on the Recommending and Granting of a Pardon’, 22 Law and Policy (2000) 49–65, Tracey E. George and Lee Epstein ‘On the Nature of Supreme Court Decision Making’, 86 American Political Science Review (1992) 323–37, Jeff Yates, ‘Presidential Bureaucratic Power and Supreme Court Voting’ 21 Political Behavior (1999) 349–66.
17 George Priest and Benjamin Klein ‘The Selection of Disputes for Litigation’, 13 Journal of Legal Studies, (1984) 1–55.
18 Estimates include 77% for the GATT [Robert E. Hudec, Enforcing International Trade Law: The Evolution of the Modern GATT Legal System, (Butterworths Legal Publishers, 1993)], and 88% for the WTO [Keisuke Iida ‘Why Does the World Trade Organization Appear Neoliberal? The Puzzle of the High Incidence of Guilty Verdicts in WTO Adjudication’, 23 Journal of Public Policy (2003) 1–21]; this study calculates similar figures, using different techniques; see below Section III.
21 Marc L. Busch ‘Democracy, Consultation, and the Paneling of Disputes under the GATT’, 44 Journal of Conflict Resolution (2000) 425–46.
22 Garrett and McCall-Smith (1999), Steinberg (2004).
23 See WTO Members Warn Appellate Body on Amicus Procedures, Inside US Trade, 1 December 2000, as quoted in Charnovitz (2002).
24 Marc L. Busch and Eric Reinhardt, ‘Testing International Trade Law: Empirical Studies of GATT/WTO Dispute Settlement’ in Daniel L. M. Kennedy and James D. Southwick (eds), The Political Economy of International Trade Law: Essays in Honor of Robert E Hudec (Cambridge: Cambridge University Press 2002) at 475–6.
25 Charnovitz (2002) at 240, Ehlermann (2002a) at 628.
26 DSU Art 21.5 disputes relate to the correct implementation of previous Panel or Appellate Body decisions; see WTO, Understanding on the Rules and Procedures Governing the Settlement of Disputes http://www.wto.org/english/docs_e/legal_e/28-dsu.pdf (visited 8 August 2006).
27 For more information on the range of disputes analysed in this study, please contact the authors or consult with journal's website at http://jiel.oxfordjournals.org/.
28 See Leitner and Lester (2005) at 232.
29 See European Communities – Regime for the Importation, Sale and Distribution of Bananas (EC – Bananas III) WT/DS27/R, WT/DS27/AB/R; United States – Import Prohibition of Certain Shrimp and Shrimp Products (US – Shrimp) WT/DS58/R, WT/DS58/AB/R; United States – Continued Dumping and Subsidy Offset Act of 2000 (US – Offset Act (Byrd Amendment)) WT/DS217/R, WT/DS217/AB/R; WT/DS234/R, WT/DS234/AB/R.
30 See, e.g.: Japan – Taxes on Alcoholic Beverages (Japan – Alcoholic Beverages II) WT/DS8/R WT/DS10/R WT/DS11/R, WT/DS8/AB/R WT/DS10/AB/R WT/DS11/AB/R.
31 This method of classification is adopted by several writers – see Busch & Reinhardt (2002) at 460.
32 The capacity of states to influence others through diplomatic or institutional avenues. For example, states may be able to pressurize Panels and the Appellate Body through their influence over other Member States in WTO negotiations, where the rules for the DSM are set (Charnovitz 2002; Steinberg 2004); they may also have greater influence over Panel selection (Andrew W. Shoyer, ‘Panel Selection in WTO Dispute Settlement Proceedings’, 6 JIEL (2003) at 203ff; Terence P. Stewart and Mara M. Burr, ‘The WTO Panel Process: An Evaluation of the First Three Years’, 32 International Lawyer (1998) 709–35), or over reappointment of Appellate Body members, and have greater capacity to damage the credibility of DSM rulings by successfully defying sanctions (Steinberg 2004 at 249).
33 The bargaining power of individual states may be closely linked to their economic strength in terms of trade volumes and economic size. Equally, the strength of a state's national economy will affect its capacity to weather the impact of retaliatory sanctions. There are also criticisms that the WTO system is inherently biased in favour of richer states, due to its institutional structure and system of rules (See e.g.: Raghavan 2000).
34 Individual characteristics of states which, for systemic or institutional reasons, may give such states an advantage in dispute settlement proceedings, including the level of experience which a state has in dispute settlement, the practical resources available to a state in terms of finances and personnel, or simply the position of the individual state in the multilateral trading system.
35 Data were collected from: International Monetary Fund, Direction of Trade Statistics (Washington, DC: IMF); International Monetary Fund, International Financial Statistics (Washington, DC: IMF); World Bank, World Development Indicators (Washington, DC: World Bank).
36 The mean figure was calculated using only those years for which sufficient reliable data were available for all significant states in the analysis; that is, data for the years 1996–2003 were summed, and divided by 8 to achieve the mean average.
37 Under this calculation, the United States was given an Index figure of 5,269.9; Chad was given an index figure of 1.0. The small number of developing states for which substantial data were unavailable were given a nominal index figure of 0.9. This scheme was designed to standardize the economic input variables, in a way which would reflect the magnitude of the disparities in economic strength and political power between states.
38 All dispute settlement reports are available on the WTO website, at www.wto.org.
39 This aspect of the coding has been questioned, and further explanation is warranted. Procedural decisions denying claims are awarded to the Respondent on the reasoning that such decisions (essentially ‘deciding not to decide’) on a particular issue have the effect of neutralising the legal challenge to a particular measure on the relevant point. In a hypothetical dispute in which only one claim is made, and the Panel/Appellate Body exercises judicial economy and refuses to consider it could only be realistically considered a failure for the Complainant and a success for the Respondent.
40 The quantitative literature on judicial decision-making at the domestic level often makes use of logit (or probit) regression to analyse factors which influence decisions, e.g.: Iida (2003) and these techniques were adopted here. An experimental logit regression was carried out using the binary coding of results outlined above; in order to provide further detail, the percentage result variable (the percentage of arguments won by the Complainant) was used as the output for an Ordinary Least Squares regression.
41 In order to analyse the data in more detail, a secondary level of analysis was performed, using (i) a percentage score of arguments won by the Complainant, from 0 to 100 and (ii) an ordinal data set, consisting of coding each result with a value between 0 and 4, representing a spectrum of results between complete success for the Complainant (4), substantial Complainant success (the Complainant winning on three or more times as many arguments as the Respondent) (3), neutral (2), substantial Respondent success (1) and complete Respondent success (0). The results achieved by analysing these outcomes are discussed briefly.
43 Eric Reinhardt, ‘Adjudication Without Enforcement in GATT Disputes’, 45 Journal of Conflict Resolution (2001) 174–95.
45 In this study, the term ‘coalition’ refers to groupings of states consisting of either a Complainant or Respondent, and the third parties which support that state's arguments in a given dispute. There is not necessarily any other connection or similarity between states and supporting third parties in dispute settlement, and none is implied by this terminology.
46 Iida (2003) – counting Complainant success as a ruling in which any violation is found.
48 Full discussion of the OLS regression results have been omitted from this paper for the sake of brevity. Please contact the authors for further details if required.
49 Ie regressing one independent variable against one dependent, rather than many against one. This lacks the advantage of multivariate analysis (that each variable will be controlled for the influence of the others), but such analysis is not possible in this instance.
50 0.05 significance is the level at which it is possible to say that there is only a 5% probability that the result is due to chance. The one exception, Difference in Previous Use, is discussed further in Part C of this Section.
51 Further details of results are available from the author on request.
52 The variables were found not to covary unacceptably.
54 Seen in the light of the work of Busch and Reinhardt (2006), the result with respect to third parties certainly merits further investigation.
56 See, e.g.: Garrett and McCall-Smith (1999), Steinberg (2004).
57 See, e.g.: Edward Kwakwa, ‘Regulating the International Economy: What Role for the State?’ in Michael Byers (ed.), The Role of Law in International Politics, (Oxford: Oxford University Press 1999), Raghavan (2000).
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||