Authors: Clint Peinhardt and Banks P. Miller
Publication: Journal of International Economic Law, jgag010
Original source: jiel_dissenting_career_advancement_2026.pdf
Conversion note: this Markdown was generated by text extraction from the hosted file. Review equations, tables, notes, and citations before treating it as a final web copy.
Dissenting for career advancement: how Arbitrators use disagreement to increase their chances of appointment Clint Peinhardt * , , Banks P. Miller Political Science, Public Policy and Political Economy, University of Texas at Dallas, 800 W. Campbell Rd, GR 31, Richardson, TX 75080, United States *Corresponding author. Political Science, Public Policy and Political Economy, University of Texas at Dallas, 800 W. Campbell Rd, GR 31, Richardson, TX 75080, United States. E-mail: clint.peinhardt@utdallas.edu
Investor–state Dispute Settlement (ISDS) typically involves three-person panels of arbitrators who decide conflicts between foreign investors and host governments in relatively secretive deliberations. In contrast to other forms of adjudication, arbitrators in ISDS need not consider precedent, meaning that dissents are less important for shaping future jurisprudence. Dissents are also rare, occurring in fewer than 15 per cent of cases, but if dissents do not aim to influence the trajectory of law, what purpose do they serve? Arbitrators who dissent may risk their reputation for collegiality. Dissents can, however, serve as ideological signals that help garner future employment. This article considers the incentives for dissents and their impact on the subsequent career prospects of arbitrators. Our contributions to this literature are (i) to integrate the study of dissents in ISDS with a larger literature on dissents; (ii) to expand the data to multiple sets of arbitral rules, (iii) to investigate career possibilities other than serving again as an ar - bitrator, and (iv) to distinguish dissents according to which party they favour. Using a complete database of the known arbitrators from 1993 to 2022, we show that dissents improve the probability of reappointment as an arbitrator, but they reduce the probability of appointment as a chair/president of the panel and even as an attorney for the claimant. We argue that dissents serve to advertise partisan affinity in ISDS but at some cost to the career paths available to dissenters. © The Author(s) 2026. Published by Oxford University Press. All rights reserved. For commercial re-use, please contact reprints@oup.com for reprints and translation rights for reprints. All other permissions can be obtained through our RightsLink service via the Permissions link on the article page on our site—for further information please contact journals.permissions@oup.com. Journal of International Economic Law, 2026, jgag010 https://doi.org/10.1093/jiel/jgag010 Original Article Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
Investor–State Dispute Settlement (ISDS) proceedings are typically conducted by three- member arbitral tribunals tasked with resolving disputes between foreign investors and host states. The outcomes of arbitration proceedings can be kept secret, 1 but they are often unanimous with all three arbitrators agreeing with the legal reasoning and award. A minority of outcomes result in separate opinions, when one or more arbitrators decide to provide their own reasoning. Although these can express agreement with the outcomes of the tribunal, the most interesting separate opinions are dissents, which ‘express substantive disagreement with colleagues’. 2 Unlike judicial systems where separate opinions may impact future legal reasoning, the rulings of tribunals in ISDS are non-binding and dissenting opinions are relatively rare, appearing in fewer than 15 per cent of cases. Given their limited formal influence, why do arbitrators ever dissent? Dissenting opinions in ISDS may violate norms of confidentiality and collegiality, potentially straining professional relationships within the tribunal. 3 In fact, one recent paper finds that arbitrators tend to vote against their prior reputations, perhaps in order to maintain a more neutral stance. 4 Yet, dissents may serve strategic purposes, enhancing an arbitrator’s visibility and appeal for future appointments. This study investigates how dissenting opinions in ISDS function as strategic signals and asks to what extent issuing a first dissent shapes an arbitrator’s subsequent career trajectory. Drawing on a comprehensive dataset of all known ISDS arbitrators through 2022, we ex - amine potential functions of dissent and its impact on future appointments. We apply insights from the political science literature on judicial behaviour to better understand the stakes of dissent and its strategic value. Our findings suggest that dissents can serve as valuable signals to potential future employers, increasing the likelihood of appointment by one of the disputing parties (claimant or respondent), but at a cost, decreasing the likelihood of selection as tribunal president or service as an attorney or expert witness. In a system where precedent holds limited sway, dissents may function less as expressions of legal disagreement and more as tools for professional positioning. When the ‘long reach’ of dissent is diminished, its value may lie primarily in strategic signalling rather than jurisprudential influence. Although dissenting behaviour has been extensively studied in domestic judicial contexts, its role in international investment arbitration remains underexplored. A large legal literature explores the incentives of arbitrators in depth, but often without systematic em - pirical evidence. 5 Todd Tucker’s Judge Knot introduces the concept of partisan wingmen and draws on interviews with arbitrators. 6 In an edited volume from the PluriCourts 1 Emilie M Hafner-Burton, Zachary C Steinert-Threlkeld and David G Victor, ‘Predictability Versus Flexibility: Secrecy in International Investment Arbitration’ (2016) 68 World Politics 413. For an argument that arbitration is increasingly transparent, see Esm �e Shirlow, ‘Transparency in Investment Treaty Arbitration: Past, Present, and Future’ (2025) 16 Journal of International Dispute Settlement 19. 2 Susan D Franck, Arbitration Costs: Myths and Realities in Investment Treaty Arbitration (OUP 2019) 72. 3 Albert Jan van den Berg, ‘Dissenting Opinions by Party-Appointed Arbitrators in Investment Arbitration’ in M Arsanjani, Looking to the Future: Essays on International Law in Honor of W. Michael Reisman (Martinus Nijhoff Publishers; Brill Academic Publishers 2011); Patricia Jimenez Kwast, ‘Prohibitions on Dissenting Opinions in International Arbitration’ What’s Wrong with International Law? Liber Amicorum A.H.A. Soons (2015). 4 Weijia Rao, ‘Are Arbitrators Biased in ICSID Arbitration? A Dynamic Perspective’ (2021) 66 International Review of Law and Economics 1. 5 Important exceptions include Franck (n 2) and van den Berg (n 3). 6 Todd N Tucker, Judge Knot: Politics and Development in International Investment Law (Anthem Press 2018). 2 • Journal of International Economic Law, 2026, Vol. 00, Issue 00 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
project, Kapeliuk examines the relationship between dissent and legal origin, finding no significant correlation. 7 Sergio Puig and Anton Strezhnev have contributed a series of papers analysing the institutional environment and incentives facing ISDS arbitrators, in - cluding dissenting behaviour. 8 Our study builds on and extends this literature in four key ways. First, we integrate theories of judicial behaviour from American politics and international organizations into the study of ISDS dissents, highlighting strategic parallels despite institutional differences. Secondly, our dataset spans a broader range of institutional settings, not just cases from the International Center for the Settlement of Investment Disputes (ICSID). Thirdly, whereas most previous work considers only reappointment as arbitrator, we assess a wider array of career outcomes, including subsequent participation as attorneys or experts. Fourthly, we categorize dissents by the party they favour and by their substance, allowing us to understand potential effects of dissent more precisely. The next section reviews the broader literature on dissenting behaviour. We then introduce the institutional context of ISDS and outline hypotheses regarding the career implications of dissent and the evolving composition of arbitral tribunals. Following this, we present our research design and empirical findings. We conclude by discussing limitations and directions for future research. Dissents and judicial behaviour Dissents are one category of separate opinions, which, as Franck highlights, can include ‘both elements of agreement and disagreement in reasoning, issues, or outcomes’. 9 Many previous studies have examined the impact of dissents, especially in American judicial be - haviour. The relative rarity of dissents suggests that they might be a potent signal, but judicial threats of dissent can alter judicial outcomes even when they are not enacted. 10 As a result, the study of formal dissents is bound to be incomplete. 11 Indeed, some evidence indicates that judges on the federal courts of appeals dissent strategically, in order to signal a reviewing court that a decision is misaligned ideologically, sometimes referred to as whistleblowing dissents. 12 Signalling is also a reason for dissent at the U.S. Supreme Court, 7 Daphna Kapeliuk, ‘Dissents in Investment Arbitration: On Collegiality and Individualism’ in Daniel Behn, Ole Kristian Fauchald and Malcolm Langford, The Legitimacy of Investment Arbitration: Empirical Perspectives (CUP 2022). 8 Sergio Puig, ‘Social Capital in the Arbitration Market’ (2014) 25 European Journal of International Law 387; Sergio Puig and Anton Strezhnev, ‘Affiliation Bias in Arbitration: An Experimental Approach’ (2017) 46 The Journal of Legal Studies 371; Sergio Puig and Anton Strezhnev, ‘Testing Cognitive Bias: Experimental Approaches and Investment Arbitration’ in Daniel Behn, Ole Kristian Fauchald and Malcolm Langford, The Legitimacy of Investment Arbitration: Empirical Perspectives (CUP 2022); Anton Strezhnev, ‘You Only Dissent Once: Re-Appointment and Legal Practices in Investment Arbitration’. 9 Franck (n 2) 70. We include as dissents some cases that Franck does not, as we did not feel as comfortable judging the substance of agreements/disagreements. Our overinclusive coding of dissents makes for a higher threshold for detecting career impacts. 10 The rarity of dissent is also seen as a by-product of institutional leadership on a court (e.g. Cass R Sunstein, ‘Unanimity and Disagreement on the Supreme Court’ (2014) 100 Cornell Law Review 769), meaning that future work might investigate whether, for instance, specific panel presidents are better at suppressing dissent than others. 11 Lee Epstein, William M Landes and Richard A Posner, ‘Why (And When) Judges Dissent: A Theoretical And Empirical Analysis’ (2011) 3 Journal of Legal Analysis 101. 12 Deborah Beim and Jonathan P Kastellec, ‘The Interplay of Ideological Diversity, Dissents, and Discretionary Review in the Judicial Hierarchy: Evidence from Death Penalty Cases’ (2014) 76 The Journal of Politics 1074. Journal of International Economic Law, 2026, Vol. 00, Issue 00 • 3 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
which has been shown to use dissent to signal to future litigants how to reframe both facts and doctrine to be more successful in future litigation, 13 something that has also been shown to be true for concurrences at the Court. 14 However, in instances where judicial scholars can account for some degrees of uncertainty in the preferences of the potential reviewing authority, the best explanation for dissent becomes simple ideological disagreement with the opinion writer, not strategic signalling. 15 Dissents are often suppressed, and dissents are correlated both negatively with the caseload of a circuit and positively with ideological heterogeneity on a circuit. In the World Trade Organization, the rarity of dissent is seen as evidence of institutional-level suppression. 16 Dissents in the international context may also reduce eventual compliance with a decision. 17 At the International Court of Justice, dissents are prevalent, owing in part to the justices’ renewable terms, which increase their accountability to home states. 18 At the European Court of Human Rights, dissents are also common, but other courts, notably the European Court of Justice, do not issue dissenting opinions. Scholarship has settled on reappointment concerns as driving the consensus norms in the court. 19 More generally, the literature argues that judges do not dissent only to signal ideological misalignment but also to reflect genuinely held beliefs about what the law says or requires, to drive greater deliberation on a particular topic, and/or to attempt to limit the scope of the majority opinion. Dissents are also associated with collegiality costs (the smaller the number of circuit judges, the lower the dissent rate). All these findings are consistent with a strategic utility maximization approach. 20 In addition, intermediate appellate court judges in the USA of - ten audition for appointment to the Supreme Court, and this auditioning behaviour includes an increased rate of dissent meant to appeal to potential appointers. 21 The literature on state court selection in the USA also provides some insight into dissenting behaviour when judges do not enjoy life tenure as they do in the federal system (and thus may be a better comparison for international investment arbitrators). For instance, Hall shows that elected state justices are likely to suppress their dissents in death penalty 13 Vanessa Baird and Tonja Jacobi, ‘How the Dissent Becomes the Majority: Using Federalism to Transform Coalitions in the U.s. Supreme Court’ (2009) 59 Duke Law Journal 183. 14 Thomas B Bennett and others, ‘Divide & Concur: Separate Opinions & Legal Change’ (2017) 103 Cornell Law Review 817. 15 Virginia A Hettinger, Stefanie A Lindquist and Wendy L Martinek, ‘Comparing Attitudinal and Strategic Accounts of Dissenting Behavior on the U.S. Courts of Appeals’ (2004) 48 American Journal of Political Science 123. 16 Meredith Kolsky Lewis, ‘The Lack of Dissent in WTO Dispute Settlement’ (2006) 9 Journal of International Economic Law 895. Lewis also argues that the suppression of dissents in the WTO context is harmful to the development of law. However, some scholars argue that a paucity of dissent in international courts may simply be due to judicial temperament (see Robin CA White and Iris Boussiakou, ‘Separate Opinions in the European Court of Human Rights’ (2009) 9 Human Rights Law Review 37, with respect to the European Court of Human Rights). 17 Daniel Naurin and Øyvind Stiansen, ‘The Dilemma of Dissent: Split Judicial Decisions and Compliance with Judgments from the International Human Rights Judiciary’ (2020) 53 Comparative Political Studies 959. 18 Jeffrey L Dunoff and Mark A Pollack, ‘The Road Not Taken: Comparative International Judicial Dissent j American Journal of International Law j Cambridge Core’ (2022) 116 American Journal of International Law 340. 19 See eg ibid. and Jeffrey L Dunoff and Mark A Pollack, ‘The Judicial Trilemma’ (2017) 111 American Journal of International Law 225. 20 Epstein (n 11). 21 Ryan C Black and Ryan J Owens, ‘Courting the President: How Circuit Court Judges Alter Their Behavior for Promotion to the Supreme Court’ (2016) 60 American Journal of Political Science 30. 4 • Journal of International Economic Law, 2026, Vol. 00, Issue 00 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
cases to avoid electoral backlash. 22 Szmer and coauthors find that judges who are elected are more likely to dissent than are those appointed to the bench, potentially as a way to ad - vertise their ideological bona fides to potential voters. 23 That signalling dynamic is constrained by the case’s salience—the more salient the case, the less likely dissent may be, regardless of the selection system. 24 This desire to dissent strategically vis- �a-vis those making future employment decisions is also likely to hold in the context of ISDS, but it is an open question of whether the potentially high salience of these cases suppresses dissent from arbitrators, given that the analogy of a voter to a future employer is imperfect. Investment arbitrators and ISDS dissents Many legal scholars argue that dissents can be problematic for arbitration generally be - cause they can reopen debate on the merits and reduce incentives for unanimity, while others emphasize the ability of dissents to improve the development of law and improve confidence in the process. 25 Particular controversy surrounds the ability of dissents to contribute to the development of international investment law, as arbitration panels are not bound by stare decisis, and few examples exist of dissents being cited in other awards. 26 To be reselected for future panels, we argue that arbitrators have conflicting incentives. On the one hand, a small group of individuals has held a disproportionate number of ISDS appointments. Rogers treats the ‘market for international arbitrators’ as a ‘relatively closed system that is difficult for newcomers to penetrate’. 27 Puig notes the network’s ‘small world properties’ by which he means that the number of repeat arbitrators is small, densely connected, and not subject to outside influence. 28 Puig found that arbitrators were 93 per cent male and that two women (Brigitte Stern and Gabrielle Kaufman-Kohler) made up 75 per cent of appointments of women to arbitration tribunals. Franck et al. make the case just as starkly: of 951 total appointments of arbitrators between 2012 and 2017, only three non- white women from developing countries were ever appointed. 29 A later study found similarly that ‘nearly all of the most prominent and repeatedly appointed arbitrators in ISDS cases are men from the Global North with significant experience in ISDS cases’. 30 Langford and coauthors analyse the community of arbitrators as a social network and demonstrate the importance of a small number of individuals who ‘double-hat’ or serve as arbitrators in some cases and as legal counsel or experts in others. 31 Thanks to this small, interconnected 22 Melinda Gann Hall, ‘Constituent Influence in State Supreme Courts: Conceptual Notes and a Case Study’ (1987) 49 The Journal of Politics 1117. 23 John Szmer, Robert K Christensen and Erin B Kaheny, ‘Gender, Race, and Dissensus on State Supreme Courts’ (2015) 96 Social Science Quarterly 553. 24 See, e.g., Richard P Caldarone, Brandice Canes-Wrone and Tom S Clark, ‘Partisan Labels and Democratic Accountability: An Analysis of State Supreme Court Abortion Decisions’ (2009) 71 The Journal of Politics 560. 25 van den Berg (n 3), Charles N Brower and Charles B Rosenberg, ‘The Death of the Two-Headed Nightingale: Why the Paulsson—van Den Berg Presumption That Party-Appointed Arbitrators Are Untrustworthy Is Wrongheaded’ (2013) 29 Arbitration International 7; Kwast (n 3). 26 van den Berg (n 3) and Brower and Rosenberg (n 25). 27 Catherine A Rogers, ‘The Vocation of the International Arbitrator’ (2004) 20 American University International Law Review 968. 28 Puig (n 8) 419. 29 Susan D Franck and others, ‘The Diversity Challenge: Exploring the Invisible College of International Arbitration’ (2014) 53 Columbia Journal of Transnational Law 429. 30 Andrea K Bjorklund and others, ‘The Diversity Deficit in International Investment Arbitration’ (2020) 21 The Journal of World Investment & Trade 410. 31 Malcolm Langford, Daniel Behn and Runar Hilleren Lie, ‘The Revolving Door in International Investment Arbitration’ (2017) 20 Journal of International Economic Law 301. Journal of International Economic Law, 2026, Vol. 00, Issue 00 • 5 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
group of legal specialists, getting appointed and reappointed as an arbitrator means being seen as easy to work with by other arbitrators. For convenience, we refer to this incentive as collegiality. On the other hand, arbitrators can develop reputations as ‘partisan wingpeople’ who look out for the interests of the investor or state that appointed them. 32 Strezhnev argues that such arbitrators are likely to be reselected based on their forceful advocacy of either claimants (investors) or respondents (states). 33 If all arbitrators consistently voted in fa - vour of their appointing party, then dissents should be relatively common, as claimant and respondent appointees refuse to agree. But such outcomes are rare—our data show high levels of unanimity, with 86 per cent of cases decided unanimously. Arbitrators who are too forceful or who are particularly known as partisan advocates for one side are unlikely to be approved in the future by an opposing party, or are more likely to be subject to challenges when appointed. 34 As such, arbitrators should self-censor their actions so that they are not so partisan as to be seen as illegitimate. Most arbitrators should be relatively reluctant to reveal their ‘ideology’, by which we mean their initial bias towards claimant or re - spondent before learning any facts of the case. Bishop and Reed distinguish between an arbitrator’s ‘general sympathy or predisposition’, which we view as closer to our use of ideology, and ‘positive bias or prejudice’, which they argue is ‘the most fundamental basis for disqualification’. 35 Others are less willing to name bias as a characteristic of arbitrators— one analysis of potential bias among arbitrators rules out systematic bias entirely and fo - cuses only on ‘perceived bias’. 36 Puig and Strezhnev argue that bias may be cognitive rather than ideological; they use a survey experiment of arbitration professionals and find evidence for affiliation effects, in which arbitrators tend to favour their appointing party regardless of their prior beliefs. 37 At its most straightforward, the theory of partisan wingpeople suggests that arbitrators will occasionally dissent to signal their ideological inclinations to future employers/ appointers, and for ISDS, that is often focused on future claimants or their law firms. But future employment can take several forms. The most common theory in the social science literature is that arbitrators want to be reappointed as arbitrators. 38 But the phenomenon of partisan signalling also has other implications for reappointment. Highly partisan individuals may be more likely to be chosen by the side towards which they lean (claimant/re - spondent) and are unlikely to be regularly chosen as presiding arbitrator due to their perceived bias. After all, in cases with two partisan wingpeople, the presiding arbitrator will determine the outcome of the case. Only arbitrators that are highly respected in the ar - bitration community and have a history of voting with other arbitrators should be sought after as presiding arbitrators, who are most commonly chosen by the arbitrators for the two parties. 32 Tucker (n 6). 33 Strezhnev (n 8). 34 Note the empirical implication here, that previous dissenters are more likely to be challenged by opposing parties in the future. We hope to test this in a future paper. 35 Doak Bishop and Lucy Reed, ‘Practical Guidelines for Interviewing, Selecting and Challenging Party- Appointed Arbitrators in International Commercial Arbitration’ (1998) 14 Arbitration International 395. P. 396 Note that this suggests dissents might be correlated with either challenges to arbitrator appointments or appeals. 36 Gus Van Harten, ‘Perceived Bias in International Treaty Arbitration’ in Michael Waibel and others (eds), The Backlash against Investment Arbitration (Wolters Kluwer Law & Business 2010). 37 Puig and Strezhnev, ‘Testing Cognitive Bias’ (n 8), Puig and Strezhnev, ‘Affiliation Bias in Arbitration’ (n 8). 38 Strezhnev (n 8). 6 • Journal of International Economic Law, 2026, Vol. 00, Issue 00 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
Reappointment as an arbitrator may not even be the best outcome for dissenters. Puig suggests that private practice can be more lucrative. 39 He notes that while ICSID arbitrators made on average $200,000 per case, some ‘complain about the low rates of pay’ relative to other venues (another venue, the London Court of International Arbitration, paid around twice that number). 40 Puig also noted that some lawyers with a ‘private law background consider ICSID ‘pro bono 0 and refuse to take many cases’. 41 The concept of reappointment as arbitrators is therefore too narrow to gauge the full impacts of dissents, which may lead not to more appointments as arbitrators but instead to paid work as counsel or as an expert across multiple cases. Ultimately, revealing ideology can also impact future work negatively. Some experienced arbitrators argue in print that appointing partisan wingpeople can backfire. Bishop and Reed use the phrase ‘advocate arbitrator’ as someone who promotes their appointing party’s perspective at the expense of neutrality, and they argue that other arbitrators will discount the opinion of such partisans. 42 Lopez argues based on his experience that ‘the ap - pointment of partial or dependent arbitrators is counterproductive as the chairman … will soon realize the fact, and the influence of this arbitrator in the deliberations will be greatly reduced’. 43 A decade later, Legum and Motin echo those sentiments: appointing a partisan ‘is neither realistic nor wise’ for two reasons. 44 First, decisions really are based on evidence and law, and second, a partisan wingperson can alienate the other two arbitrators. 45 It is striking that the practitioner literature leans so heavily in the direction of collegiality. In this line of thinking, dissents are costly to individuals and may serve as retrospective judgments on the interpersonal dynamics of a given tribunal’s experience. These authors would clearly expect dissents to reduce future employment as an arbitrator, and perhaps to affect other types of employment as well. The signalling value of dissents is therefore not straightforward and is conditioned by competing incentives. Arbitrators who want repeat business as advocates of their appointing party may use dissents to signal their ideologies, but doing so too often may result in fewer appointments. Even the most partisan of arbitrators should use dissents strategically when considering their career paths, but ultimately, we believe that signalling partisanship can be professionally rewarding. Hypothesis 1 reflects the logic of partisan wingpeople and using dissents as partisan signals. H1: Arbitrators who dissent are more likely to be reappointed by parties (either claimant or re - spondent) in the future While our first hypothesis reflects the benefits of signalling, our second hypothesis reflects its cost. To be appointed as the chair or president of a tribunal, a reputation for collegiality and for considering the facts of the case and the applicable law should be paramount. We therefore expect to observe one cost of collegiality in the number of times previous dissenters are appointed as presiding arbitrator. 39 Puig (n 8). 40 ibid. 398. 41 ibid. 42 Bishop and Reed (n 35). 43 Carlos Alberto Matheus Lopez, ‘Practical Criteria for Selecting International Arbitrators’ (2014) 31 Journal of International Arbitration 795. 44 Barton Legum and Clara Motin, ‘The Essential Qualities for an Arbitrator’ (2023) 38 ICSID Review - Foreign Investment Law Journal 42-3. 45 Many others echo this point, such as Brower and Rosenberg (n 25) 15.j Journal of International Economic Law, 2026, Vol. 00, Issue 00 • 7 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
H2: Arbitrators who dissent are less likely to be appointed as a presiding arbitrator in the future Examining how a dissenting opinion impacts career pathways assumes that all dissents reveal the same information, which they may not. Any dissent may reduce the likelihood of a future appointment as presiding arbitrator, but if investment arbitration is really a small world network, then the direction of the dissent may matter as well. Dissents in fa - vour of the claimant (investor) should result in more work for claimants, but less for respondents. That work could mean reappointment by other claimants as an arbitrator, or it could take the form of other legal work. Dissents in favour of respondents should correspondingly result in more legal work for respondents, perhaps as arbitrator but also as counsel or expert witness. The following hypotheses make these expectations explicit. H3(a): Dissents favouring claimants should result in more appointments by claimants as an arbitrator. H3(b): Dissents favouring claimants should result in more legal work for claimants. H3(c): Dissents favouring respondents should result in more appointments by respondents as an arbitrator. H3(d): Dissents favouring respondents should result in more legal work for claimants. Note that by ‘more legal work’ we mean either representing clients in arbitration settings or providing expert opinions in arbitration settings. Research design In order to test these hypotheses, we gather dispute (case) and arbitrator data from the Investor State Law Guide, which includes information not just on tribunal membership but also on legal representation and expert testimony. 46 We only include cases that have a decision, so we eliminate those that are settled or that are still pending. We then reassemble data at the arbitrator level and code dissenting opinions. In theory, we are interested in dissenting votes, but generally we have access only to dissenting opinions. 47 We focus only on arbitrators’ first dissent, for two reasons. First, most arbitrators never dissent, and most who do dissent only do so once (n = 68). Only fifteen arbitrators dissent twice or more, and only seven dissent more than three times. One arbitrator, Charles Brower, is credited with ten dissents in our data. Were we to include multiple dissents, he would be a clear outlier, and we would conflate subsequent appointments and dissents. 48 The arbitrators with the second highest total (five) are Philippe Sands and Brigitte Stern, who are well known in the ISDS community. Ultimately, we examine only the first dissent because one dissent represents the modal number of dissents and because we feel that dissents beyond the first provide a diminishing effectiveness as a signal. Secondly, using the first dissent allows us to assess the effects of dissents across a career and many potential subsequent appearances. These data include 4472 appointments across six role types: president of the panel, claimant-appointed arbitrator, respondent-appointed arbitrator, claimant attorney, respondent attorney, and expert. In total, we track the careers of 472 unique arbitrators. 46 Data were downloaded in summer of 2024. https://www.investorstatelawguide.com 47 Where no opinion is available, we still count dissents in our Table 2 results, but not in the directional results ( Table 3 ). 48 Given Brower’s frequent appearances in our data, we examined our results when dropping him from the sample. Results remain unchanged. 8 • Journal of International Economic Law, 2026, Vol. 00, Issue 00 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
Our dependent variable in the analyses that follow is the role occupied by someone who served as an arbitrator in a subsequent case. As mentioned above, there are six potential roles that an arbitrator can assume in our data: claimant-appointed arbitrator (coded 1), a respondent-appointed arbitrator (coded as 2), the president of a panel (coded 3), an attorney for the claimant (coded as 4), an attorney for the respondent (coded 5), or an expert (coded 6). In any given case, these categories are mutually exclusive. We collected several variables beyond our dependent variable of role type. Our key independent variable is whether an arbitrator dissented (post-1st dissent). This variable is coded one after the first dissent and zero before. To give some intuition on the coding for this variable, assume that arbitrator A hears four cases in the data. If she dissents in the second case, then we count appointments in cases 3 and 4 for the purposes of our dependent variable. About 17 per cent of the appointments in our data occur after an initial dissent by an arbitrator. From biographical information scraped from the web, we coded whether an arbitrator had an academic background (academic position; 1 if yes, 0 otherwise), meaning that this individual had a connection to an institution of higher education. Second, we coded whether an arbitrator held a position in a government (government position; 1 if yes, 0 otherwise). We coded gender based on arbitrator name (woman; 1 if a woman and 0 otherwise). We coded World Bank classification (WB class) based on the country of residence for the arbitrator, with higher scores indicating residence in a wealthier country. This variable follows the World Bank coding scheme and ranges from 0 to 3. To control for the legal rules under which arbitration occurs, we include an indicator for whether the case was heard using ICSID rules (1 if yes, 0 otherwise)—about two-thirds of all cases are heard us - ing ICSID rules. 49 Finally, we include a yearly cubic spline with three knots to help control for any unobserved changes over time in our data (year 1–year 4).
Table 1 contains descriptive statistics for variables included in the models that follow. Our
data include all ISDS cases completed by 2022. A few things are notable in examining the de - scriptive statistics. Recall that Strezhnev (2017) finds that about 20 per cent of cases include a dissent. Our data, which extends past his 2015 end date and which include non-ICSID disputes, show a lower rate of dissent. We estimate that 14 per cent of cases in our data have a dissent. Arbitrators in ISDS The rate of dissent holds relatively steady until 2016, after which it declines precipitously.
Figure 1 plots the dissent rate, total number of dissents, and a smoothed line for the dissent
rate over time. The x-axis displays the date a case was initiated. It is apparent that there is a sharp decline in the dissent rate and in the number of dissents starting in 2016. This de - cline is not explained by change in the governing bodies under which these arbitrations oc - cur, as the percentage of ICSID cases remains more or less constant between 50 per cent and 70 per cent of all resolved cases. This is also not likely to be the result of a diversifying pool of arbitrators over time, as we explore in Appendix A . Another alternative is that this is an artefact of our ability to collect data. As a reminder, our data include only cases completed by 2022, and it may be that cases that generate dissent also take longer to reach a conclusion. If this is so, then the decline in dissent rate that we observe post-2015 is simply a selection effect: those cases in which dissent occurs are unlikely to have been completed 49 For reasoning on why this may be important see Stephen Jagusch and Jeffrey Sullivan, ‘A Comparison of ICSID and UNCITRAL Arbitration: Areas of Divergence and Concern’ in Michael Waibel and others (eds), The Backlash against Investment Arbitration (Wolters Kluwer Law & Business 2010). Journal of International Economic Law, 2026, Vol. 00, Issue 00 • 9 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
in 7 years between 2015 and 2022. Yet this does not fully explain why the drop appears suddenly in the data, rather than exhibiting a more gradual decline. Results We now focus on our primary concern, a more granular analysis of the career patterns of individual arbitrators and what it reveals about the impact of dissent on career prospects. We note at the outset that our results reflect correlations among variables and cannot be interpreted as strictly causal, but it does reflect a time order that is consistent with our theory. To test our first two hypotheses, we use a multinomial logit model in which there are six potential positions to which an individual might be appointed: president of the panel, a
Variable Mean Std. dev. Min. Max. Role 2.86 1.48 1 6 Post-1st dissent 0.17 0.38 0 1 Gov. position 0.36 0.48 0 1 Academic position 0.70 0.46 0 1 Woman 0.13 0.34 0 1 World Bank class. 2.82 0.44 0 3 ICSID 0.65 0.48 0 1 Year 1 2011 6 1993 2022 Year 2 6.09 5.94 0 21 Year 3 1.46 1.88 0 6.93 Year 4 0.13 0.23 0 0.96
10 • Journal of International Economic Law, 2026, Vol. 00, Issue 00 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
claimant-appointed arbitrator, a respondent-appointed arbitrator, an attorney for the claimant, an attorney for the respondent, or an expert. Including all these potential options is crucial because it may be the case that dissent has uneven effects on career prospects, si - multaneously raising some prospects while lowering others. We focus on what occurs after an initial dissent by an arbitrator. Figure 2 presents a coefficient plot of the change in predicted probability for each of the six subsequent positions. In this coefficient plot, a dot represents the point estimate, and a line shows the 95% confidence interval around that point estimate. We find that initial dissents predict a significant subsequent increase in serving as an arbitrator for both the claimant and the respondent in these cases. These results support our first hypothesis. Dissent increases the likelihood of serving as an arbitrator for both claimants and respondents, but the magnitude of the in - crease is asymmetric. Respondent arbitrators are more likely to have dissented in the past than are claimant arbitrators (28 percentage points versus 11 percentage points).
arbitrator is offset by statistically significant decreases in the likelihood of service as a panel president or as an attorney for claimants or respondents. Given that the fees for service as an attorney may be substantially higher than those for service as an arbitrator, this is not a trivial tradeoff and may work to suppress dissent. These initial results suggest that there are more downsides to dissent for those who are initially appointed as arbitrators by claimants, as the potential negative impacts of dissent are larger for those on the claimant side. Dissent does less to promote the odds of becoming an arbitrator appointed by the claimant and more to reduce the odds of service as an attorney for the claimant. Conversely, dissent greatly increases the odds of service as a respondent-appointed arbitrator and less to reduce the odds of appointment as attorney for the respondent. And as Traxler notes, the processes of appointment are not symmetric for
Journal of International Economic Law, 2026, Vol. 00, Issue 00 • 11 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
multiple reasons—claimants have at least a time advantage for choosing their arbitrator and can choose more carefully. 50 Perhaps that translates to an easier time coordinating, but the finding raises more questions than it answers, and a more thorough investigation of the role that law firms play in choosing arbitrators is worthwhile. Much of the previous scholarship on arbitrators focuses on their demographic characteristics, but our behavioural measure of dissent appears to be a better predictor of future employment.
Table 2 presents a complete picture of how the behavioural measure of dissent compares with
the primary demographic characteristics that have been so prominent in the study of arbitrators. We have bolded the statistically significant effects for variables that are correlated with various career choices. Notice that dissent is a statistically significant covariate with all career outcomes, whereas none of the demographic variables is significant for more than four. We examine country of origin, gender, and professional background. Our results in Table 3 are in keeping with previous findings. Arbitrators from wealthier countries are significantly less likely to serve as arbitrators for respondents, which is unsurprising considering the composition of countries that have been sued in the ISDS process. Arbitrators from wealthier countries are significantly more likely to serve as presidents of panels and as attorneys for claimants. We also replicate the male dominance of the arbitration profession: women are much less likely to be appointed as arbitrators by claimants. Examining non-arbitrator in - volvement in cases surfaces a new result, however, as women are more likely than men to be hired as attorneys for respondents. Professional backgrounds matter in different ways. Those who work for governments (or have in the past) are more likely to be appointed as presidents of panels but are less likely to serve as claimant attorneys. Arbitrators with academic backgrounds are more likely to serve as arbitrators appointed by respondents and as experts, but are, like those with government service, much less likely to be hired as counsel for claimants. Together, these results highlight the extent to which claimants seem to have a tight focus on who they patronize with their legal business. Those with backgrounds in government or academia are less likely to be hired, while those from rich countries are more likely to be hired. This contrasts with what looks to be relatively less strategic hiring practices for respondents. One limitation of the analysis thus far is that it assumes that all dissents have the same effects, but the community of investment arbitrators is still small, and an entire industry supports the dissemination of case law to legal insiders. We therefore argue that dissents should be widely publicized and that their use as partisan signals may be particularly effective in ISDS. To check the robustness of our results, we code every dissent in our data for ‘directionality’, or whether it on balance favours the claimant or the respondent more. Arbitrators appointed by claimants dissenting for the first time do so in favour of claimants 70 per cent of the time, while those appointed by respondents dissent in favour of respondents 76 per cent of the time in our data. 51
rectional dissents, and in general, the results are like those described in Figure 2 and
depends on whether the dissent is in favour of the claimant (left side) or the respondent (right side). The most important takeaways from Figure 3 are that pro-claimant dissents in - crease the subsequent likelihood of being appointed as an arbitrator by the claimant by 19 percentage points on average, and that pro-respondent dissents increase the likelihood of 50 Tobias Traxler, ‘Pro-Claimant Bias in Arbitrator Selection’ (2024) 27 Journal of International Economic Law 54. 51 Note that our estimates differ with van den Berg (n 3), who finds that arbitrators always dissent in favor of the appointing party. These differences may be due to the sample of cases or to interpretation/coding. 12 • Journal of International Economic Law, 2026, Vol. 00, Issue 00 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
Table 2 Full substantive results. Boldface numbers indicate statistical significance (at a 95% level of confidence).,.
Role Dissent Government Academic Woman WB Class. Claim. Arbitrator 11 [1, 22] a −6 [−13, 1] −13 [−22, −5] −16 [−22, −10] 10 [−3, 23] Resp. Arbitrator 28 [13, 42] 3 [−5, 11] 14 [7, 21] 9 [−7, 24] −19 [−34, 4] President −17 [−23, −10] 11 [3, 19] 2 [−6, 9] 3 [−11, 17] 13 [3, 23] Claim. Attorney −10 [−14, −6] −6 [−11, −1] −6 [−11, −1] 0 [−7, 7] 7 [1, 14] Resp. Attorney −8 [−14, −2] 0 [−8, 7] −2 [−10, 6] 9 [−2, 19] −13 [−30, 4] Expert −4 [−6, −1] 0 [−5, 3] 5 [2, 7] −4 [−7, −2] 2 [−1, 6] a Note: Boldface numbers indicate statistical significance at p<0.05, two-tailed (with 95% confidence intervals in brackets). Journal of International Economic Law, 2026, Vol. 00, Issue 00 • 13 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
subsequent appointment by the respondent as an arbitrator by 27 percentage points on av - erage. In other words, adding directionality of the dissent shows our best evidence yet of dissent as signalling for partisan wingpeople (H1). Furthermore, the partisan dynamic works for both claimant-appointed (H3a) and respondent-appointed (H3c) arbitrators, and the magnitude of the effect increases substantially for pro-claimant dissents. Likewise, the directional results continue to show that dissents of any type lower future chances of being a presiding arbitrator (H2).
Figure 3 replicates our Figure 2 finding that both pro-claimant and pro-respondent dissents
decrease the likelihood of serving as an attorney for claimants, but pro-respondent dissents clearly have a bigger negative effect. Importantly, these results contradict our initial assessments that respondent arbitrators may suffer fewer costs for dissents. The directional results allow us to reject Hypothesis 3b, as our expected result is that dissents favouring claimants should increase the chances of representing a claimant as an attorney. We also find no evidence for hypothesis 3d, as dissents favouring respondents have no effect on working as an attorney for respondents. This asymmetry is a bit surprising and shows that the career impact of dissents is not as straightforward for non-arbitrator tracks. Relatedly, dissents favouring claimants reduce future appearances as an expert, but those favouring respondents have no such effect. Full results for this model are presented in Appendix A . One last concern is that we have thus far not distinguished the number of dissents. Broadly speaking, we can sub-classify dissents into three non-mutually exclusive categories: dissents based on jurisdiction, dissents based on the merits, and dissents based on the quantum or
Jurisdiction Merits Damages Resp. arb. 0.30 0.65 0.19 Claim. arb. 0.29 0.59 0.33
14 • Journal of International Economic Law, 2026, Vol. 00, Issue 00 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
damage awarded. We coded every dissent in our data into these categories, with the caveat that whether something is a merits or jurisdictional dissent can occasionally be hard to distinguish. In our data, 61 per cent of the dissents involve questions of merit, 31 per cent questions of jurisdiction, and 24 per cent questions of damages. In addition, 12 dissents involved issues of both merits and damages, seven involved both jurisdiction and merits, three included both jurisdiction and damages, and 1 involved all three areas. Table 3 describes the dissents in our data by appointing party of the arbitrator and type of dissent. These do not sum to one in each row be - cause dissents can include multiple elements that cross categories. The results show that claimant-appointed arbitrators focus more on damages than respondent-appointed arbitrators and that most dissents have a merits component. Ultimately, most dissents are about something substantive in the case, which shows that (as we expect) collegiality and legal reasoning are im - portant constraints on partisan signalling. Second, we can look at the raw numbers of first dissents in each category by direction, that is favouring either the respondent or the claimant. Note, the total is 142, which is more than the total number of dissents because of the non-exclusivity of the categories. This tells the same basic story as Table 3 . Lastly, we can look at what later positions were taken by arbitrators who dissented in a particular direction, by the substance of the dissent. Table 5 shows that no pro-respondent dissents on jurisdiction subsequently ended up serving as arbitrators for claimants, five served as respondent arbitrators, and only one served as a panel president. Again, we see that one cost of dissent dramatically lowers the chance of being appointed as a presiding arbitrator in the future. Tables 3–5 show that the types of dissent do not differ dramatically across the source of appointment, providing confidence that collegiality norms—and thus the need to make
Pro-respondent 1st dissents Jurisdiction 6 Merits 20 Damages 10 Pro-claimant 1st dissents Jurisdiction 32 Merits 55 Damages 19
Respondent Jurisdiction Merits Damages Claim arb. – 5 2 Resp. arb 5 15 8 President 1 – – Claimant Jurisdiction Merits Damages Claim arb. 20 41 11 Resp. arb. 11 13 8 President 1 – – Journal of International Economic Law, 2026, Vol. 00, Issue 00 • 15 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
plausible legal arguments—–constrain the partisan dynamics that we find earlier. However, Tables 4 and 5 show that respondent-appointed arbitrators dissent less often, in - dicating that partisan signalling may disproportionately operate for claimant-appointed arbitrators. Conclusion This article investigates the strategic role of dissents in ISDS and their implications for arbitrators’ career trajectories. Drawing on a comprehensive dataset of arbitrators from 1993 to 2022, our evidence is consistent with the notion that dissents—–though rare—–serve as potent signals in a professional environment where ideological alignment and reputational considerations are paramount. Our findings demonstrate that dissenting arbitrators are more likely to be reappointed by both claimants and respondents, and that this effect is driven by the partisan direction of the dissent. These results are the strongest evidence to date that ISDS dissents function as advertisements of partisan ideology. Such signalling comes at a cost to collegiality, however, and dissenters are less likely to be appointed as presiding arbitrators or as attorneys for claimants. These results contribute to the broader literature on judicial behaviour by extending theories of strategic dissent to the international arbitration context. Unlike domestic courts, where dissents may shape precedent or reflect principled legal disagreement, ISDS dissents appear to be more closely tied to careerist incentives. The asymmetry in career outcomes, particularly the diminished prospects for dissenters to serve as claimant attorneys, suggests that investors as claimants may engage in more strategic hiring practices, while respondents may be more tolerant of ideological signalling. Overall, the article raises interesting questions about the profession of arbitration and careerist incentives for arbitrators, and highlights similarities more than differences between the behaviour of arbitrators and other adjudicators. Several limitations temper our conclusions. First, our coding for both directionality and types of disagreement is relatively coarse. Future researchers may want to refine this coding to better capture the nuances of dissent. 52 Secondly, our data only include completed cases through 2022, and reputational dynamics may change after that time, in part because we are highlighting them. Third, while we observe significant career effects, we cannot fully disentangle whether these are driven by the content of dissents, the reputations of dissenters, or the strategic behaviour of appointing parties and law firms. Some dissents are from arbitrators whose ideological leanings are probably already well known by their first dissent (eg, Charles Brower on the claimant side and Brigitte Stern on the re - spondent side). For arbitrators, these findings underscore the benefits and costs of strategic arguments. Dissenting may enhance visibility and appeal to certain stakeholders, but it also risks exclusion from roles that demand impartiality. For the ISDS system more broadly, the results raise questions about the legitimacy and transparency of arbitrator selection. If dissents function primarily as career signals rather than legal critiques, then the adjudicative process may be significantly shaped by market dynamics, which has implications for reform 52 Franck (n 2), which unfortunately we did not discover until after the bulk of our analysis was completed, provides an excellent framework for more nuanced coding. 16 • Journal of International Economic Law, 2026, Vol. 00, Issue 00 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
efforts aimed at enhancing the independence and diversity of ISDS tribunals. Understanding the career incentives behind dissent is therefore essential not only for scholars of judicial behaviour but also for policymakers and practitioners seeking to improve the fairness and credibility of international arbitration. Future work should examine more closely the impact of dissents on future challenges to the appointment of arbitrators, because if dissent increases the odds of a successful challenge, it should reduce the market dynamics we find evidence for. Separately, re - search is needed to reconcile Rao’s finding that arbitrators try to maintain a reputation for neutrality, with our results here. 53 In our own work, we hope to explore the interaction of arbitrators and law firms more closely in order to sort out some of the asymmetrical results in this study. The role of law firms in choosing arbitrators is clearly understudied, and they may be the demand side of the partisan wingpeople argument that we have found evidence for here. Additionally, the selection of presiding arbitrators deserves separate attention—in many cases, they are chosen by the two party-appointed arbitrators, but when those two cannot agree, different arbitration rules specify appointing authorities. Who gets those appointments, and how do they differ from party- appointed selections? More generally, we see additional value to consider ISDS from the perspective of the judicial behaviour literature more broadly, and we hope this article will inspire others to do so. Acknowledgements We thank Mark Pollack and panel participants from the American Political Science Association and Midwest Political Science Association meetings in 2025. We are im - mensely grateful to Aishah Abdullah, Mamie Cincotta, Spencer Ng, Leah Sullivan, and Giuseppe Peressotti for research assistance. Author contributions Banks Miller (Conceptualization [equal], Data curation [equal], Formal analysis [lead], Funding acquisition [equal], Investigation [equal], Methodology [lead], Supervision [Supporting], Visualization [lead], Writing—original draft [equal], Writing—review & editing [Supporting]) Funding We acknowledge and appreciate assistance from the Political Science Program at the University of Texas at Dallas for financial support of the research.
Below we plot the diversity of the arbitrator pool over time, using gender and other basic background characteristics. The important takeaway from Figure A1 is that none of these 53 Rao (n 4). Journal of International Economic Law, 2026, Vol. 00, Issue 00 • 17 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
characteristics exhibit a clear departure from trends around the 2015 cliff in dissent rates we note in the manuscript (which is emphasized using the vertical dashed line). On the left side of the y-axis are averages per year for gender, whether an arbitrator has an academic background, and whether an arbitrator has a government background. On the right side of the y-axis is the measure of wealth according to the World Bank classification, with three indicating a wealthy state.
Multinomial logit Claim. arbitrator Resp. arbitrator President Claim. attorney Resp. attorney Arb. characteristics Pro-resp. dissent 0.02 (.84) 1.12 (.80) −0.82 (.76) −1.35 (.91) 0.20 (.95) Pro-claim. dissent 2.56 (.47)** 2.34 (.55)** 0.98 (.50) 1.36 (.49)* 1.23 (.64) Gov. position 0.10 (.52) 0.46 (.52) 0.66 (.53) −0.36 (.54) 0.16 (.56) Academic position −2.41 (.47)** 1.06 (.51)* −1.73 (.49)** −2.22 (.38)** −1.92 (.54)** Woman 1.24 (1.16) 2.73 (1.12)* 2.28 (1.17)* 2.17 (1.10)* 2.69 (1.08)* World Bank class. −0.10 (.40) 0.61 (.36) −0.04 (.37) 0.03 (.41) −0.70 (.39) Controls ICSID rule 0.15 (.21) 0.21 (.22) 0.10 (.22) 0.32 (.23) 0.24 (.25) Year 1 −0.05 (.07) −0.04 (.07) −0.06 (.07) −0.05 (.07) −0.04 (.07) Year 2 0.44 (.36) 0.54 (.35) 0.43 (.36) 0.28 (.34) 0.09 (.36) Year 3 −1.20 (1.26) −1.75 (1.24) −1.12 (1.28) −1.09 (1.25) −0.16 (1.31) Year 4 2.50 (3.42) 4.46 (3.37) 2.17 (3.47) 3.51 (3.58) 0.11 (3.75) Constant 102.83 (145.64) 84.29 (146.62) 120.80 (147.43) 93.54 (135.97) 84.61 (144.27) N 4472 4472 4472 4472 4472 N arbitrators 473 473 473 473 473 Wald 327.56 (0.00) PRE 0.10 18 • Journal of International Economic Law, 2026, Vol. 00, Issue 00 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
Multinomial logit Claim. arbitrator Resp. arbitrator President Claim. attorney Resp. attorney Arb. characteristics Post-1st dissent 1.88 (.55) ** 2.45 (0.63) ** 0.49 (.61) 0.25 (.59) 0.67 (.66) Gov. position −0.09 (.53) 0.34 (.52) 0.66 (.54) −0.33 (.56) 0.14 (.58) Academic position −2.35 (.48) ** −0.98 (.51) −1.73 (.49) ** −2.22 (.48) ** −1.91 (.55) ** Woman 1.08 (1.14) 2.51 (1.12) * 2.26 (1.16) 2.14 (1.09) * 2.66 (1.09) * World Bank class. −0.08 (.41) −0.73 (.36) * 0.02 (.38) 0.05 (.42) −0.71 (.41) Controls ICSID rule 0.22 (.22) 0.27 (.22) 0.13 (.22) 0.35 (.23) 0.27 (.25) Year 1 −0.06 (.07) −0.06 (.07) −0.06 (.07) −0.04 (.07) −0.05 (.07) Year 2 0.36 (.35) 0.49 (.33) 0.39 (.35) 0.24 (.34) 0.09 (.35) Year 3 −0.96 (1.22) −1.73 (1.18) −1.00 (1.24) −0.92 (1.23) −0.17 (1.27) Year 4 2.19 (3.34) 4.89 (3.26) 1.78 (3.39) 2.99 (3.53) 0.11 (3.68) Constant 116.67 (135.31) 126.39 (132.95) 120.75 (143.74) 84.96 (130.27) 96.54 (140.62) N 4472 4472 4472 4472 4472 No. of arbitrators 473 473 473 473 473 Wald χ 2 270.20 (0.00) PRE 0.09 * Significant at95% level of confidence; ** Significant at 99% level of confidence (both two-tailed tests). Journal of International Economic Law, 2026, Vol. 00, Issue 00 • 19 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026
Table A1 provides results for the multinomial model that we estimated in order to generate Figure 3 in the article. This model substitutes our single indicator of whether an arbitrator dissented with a set of indicators for the directionality of the dissent—either in favour
of the respondent or the claimant.
by arbitrator. Overall, the model fits the data well, with a reduction in error across all six career choices of 9 per cent and a highly statistically significant Wald χ 2 statistic. In addition, we confirm that the assumption of the independence of irrelevant alternatives holds. However, results in this model are dependent on comparisons to the base category, here, whether one serves as an expert. The results related to academic background demonstrate this impact—scholars are more likely to serve as experts than any other role (except respondent-appointed arbitrators). © The Author(s) 2026. Published by Oxford University Press. All rights reserved. For commercial re-use, please contact reprints@oup.com for reprints and translation rights for reprints. All other permissions can be obtained through our RightsLink service via the Permissions link on the article page on our site—for further information please contact journals.permissions@oup.com. Journal of International Economic Law, 2026, 00, 1–20 https://doi.org/10.1093/jiel/jgag010 Original Article 20 • Journal of International Economic Law, 2026, Vol. 00, Issue 00 Downloaded from https://academic.oup.com/jiel/advance-article/doi/10.1093/jiel/jgag010/8692553 by University of Texas at Dallas - McDermott Library user on 03 June 2026