Pro-Investor or Pro-State Bias in Investment-Treaty Arbitration? Forthcoming Study Gives Cause for Concern

Debates about investment treaties often raise questions about fairness and independence in international investment arbitration. Some observers argue that investment arbitration offers a neutral and impartial forum in which to resolve investor-state disputes as a basis for protecting foreign-owned assets and ensuring the rule of law.[1] Others claim that the arbitration mechanism favours investors and Western capital-exporting states at the expense of respondent governments, especially in the developing world.[2] Overall, principles of fairness and independence are integral to the legitimizing role of international arbitration.[3]

A forthcoming study tested hypotheses of potential bias in investment arbitration.[4] In particular, it examined whether there was evidence to support expectations that the resolution of contested legal issues in investment treaty law would be influenced by apparent economic interests of arbitrators or the arbitration industry.

The rationale for the study’s hypotheses derived from assumptions about arbitrator incentives, based on the unique make-up of investment treaty arbitration.[5] First, the asymmetrical claims structure and absence of institutional markers of judicial independence could create incentives for arbitrators to favour the class of parties that is able to initiate use of the system.[6] Second, where they have a significant career interest in the field, arbitrators could be influenced by a need to appease actors with power or influence over specific appointment decisions or over the wider position of the relevant arbitration industry.[7]

Such assumptions have been rejected by participants in the system[8] and it is not suggested in the study that they are the only possible factors that may influence arbitrator behaviour. On the other hand, the assumptions reflect the “question of the incentives that so often operate on arbitrators – that is, of their self-interest in trying to secure and expand prospects for future arbitral appointments” and the related expectations that “[a]n arbitrator may perceive that his award is likely to have an impact on his own acceptability, that is, on the probability of his being appointed again… [o]r an award may affect the marketability of the appointing organization, on which the arbitrator depends for future referrals.”[9]

Two significant tendencies were observed. First, there was a strong tendency toward expansive resolutions of contested issues of law that enhanced the compensatory promise of the system for claimants and, in turn, the risk of liability for respondent states. The second was an accentuated tendency toward expansive resolutions where the claimant was from a Western capital-exporting state.

Outline of the study

The study was based on a systematic content analysis[10] of all publicly-available awards (i.e. decisions) dealing with jurisdictional matters in 140 known cases under investment treaties to May 2010. The awards were coded for resolutions, by each of the arbitrators, of a series of legal issues of jurisdiction that were contested more broadly in awards or secondary literature. These tend to be issues on which the text of an investment treaty is ambiguous or silent, leading to disagreements about the appropriate approach. Expansive resolutions of an issue may be said to favour claimants by expanding the authority of investment treaty tribunals and by allowing more claims to proceed.

In brief, the coded topics and issues were:

Corporate person investor – Should a claim be allowed where ownership of the investment extends through a chain of companies running from the host state to the home state via a third state? Expansive approach: yes. Restrictive approach: no.

Natural person investor – Should a claim be permissible where brought by a natural person against the only state of which the person is a citizen, or against a state of which the person is a citizen without confirmation of dominant and effective nationality? Expansive answer: yes to either of the two questions. Restrictive approach: no to either of the two questions.

Investment – Should the Fedax criteria[11] be applied to limit the concept of investment under the ICSID Convention; alternatively (regardless of whether under the ICSID Convention) should there be a requirement for an actual transfer of capital into the host state as a feature of an investment or should the concept of investment be limited to traditional categories of ownership? Expansive approach: no to any of the three questions. Restrictive approach: yes to any of the three questions.

Minority shareholder interests – Should a claim by a minority shareholder be allowed where the treaty does not permit claims by minority shareholders expressly, such as where the treaty does not mention “shares” in the definition of investment, or should it be permitted without limiting the claim to the shareholder’s interest in the value and disposition of the shares (as opposed to interests of the domestic firm itself)? Expansive approach: yes to either of the two questions. Restrictive approach: no to either of the two questions.

Permissibility of investment – Should there be an evident onus placed on the claimant (or the respondent state) to demonstrate that an investment was (or was not) affirmatively approved or was (or was not) based on corrupt practices? Expansive approach: onus on the respondent state for either issue. Restrictive approach: onus on the claimant for either issue.

Parallel claims – Should a claim be allowed in the face of a treaty-based duty to resort to local remedies that clearly was not satisfied by the claimant; in the face of a contractually-agreed dispute settlement clause relating to the same factual dispute; in the face of an actual claim, arising from the same factual dispute, brought via an alternative path of a treaty-based fork-in-road clause; or in the face of an actual claim, arising from the same factual dispute, brought via another treaty where the claim could lead to a damages award in favour of the investor? Expansive approach: yes to any of the four questions. Restrictive approach: no to any of the four questions.

Scope of most-favoured-nation treatment – Should the concept of most-favoured-nation treatment be extended to non-substantive provisions of other treaties (such as dispute settlement provisions)? Expansive approach: yes. Restrictive approach: no.

Results of the study[12]

Expansive versus restrictive issue resolutions

The results supported the hypothesis that arbitrators would tend to adopt an expansive (claimant-friendly) approach to the resolution of the coded issues.[13] Table A summarizes the variation in issue resolutions. As indicated, there were different variations and different amounts of data for each issue.

Table A: Classification of Issue Resolutions by Issue

Issue No. of issue resolutions Resolution   of issue
Expansive Restrictive
1 – corporate person investor 69 81.94% 18.06%
2 – natural person investor 6 0.00% 100.00%
3 – concept of investment 116 72.27% 27.73%
4 – minority shareholder interest 72 92.00% 8.00%
5 – permissibility of investment 27 66.67% 33.33%
6 – parallel claims 165 82.74% 17.26%
7 – scope of MFN treatment 60 50.00% 50.00%
Cumulative 515 76.09% 23.91%

The results indicated that the strong tendency in favor of an expansive approach was led by resolutions of four issues: corporate person investor; concept of investment; minority shareholder interest; and parallel claims. On one issue, the scope of MFN treatment, arbitrators were split between expansive and restrictive approaches. There was insufficient data to draw conclusions on two other issues: natural person investor and permissibility of investment.

These results provide tentative cause for concern for those who expect the system to deliver a degree of evenness[14] – between the interests of claimants and respondent states – in the resolution of jurisdictional issues, where the text of the treaty is ambiguous or silent on the issue. States have lost across a range of contested issues, sometimes overwhelmingly so, in the litigation of jurisdictional objections to investor claims. If observers expected the coded issues to be resolved restrictively, this has not been the case in practice.

Role of claimant nationality

The results also supported the hypothesis that an expansive approach would be accentuated where the claimant was a national of France, Germany, the U.K., or the U.S.[15] These countries were chosen as an approximate measure of major Western capital-exporting states. The measure was supplemented by analyses of additional groupings associated with Western capital-exporting interests.

Overall, the effect of claimant nationality for these four countries, as a group, was statistically significant. Table B outlines the probability of an issue being resolved expansively in each of the five categories of claimant nationality, when other covariates were held steady.

Table B: Effects of Claimant Nationality on the Likelihood of an Expansive Resolution

Claimant nationality Probability of expansive resolution Statistical significance
France .86 p=.005
Germany .47 p=.38
United Kingdom .95 p<.001
United States .98 p<.001
All others .69 p<.001

Overall, issue resolutions in cases brought by claimants from France, Germany, the U.K., or the U.S. were 84% more likely to be resolved expansively when compared to cases brought by claimants from other states. Besides this overall effect, the analysis indicated that country-by-country variations between claimants from France, the U.K., or the U.S. and claimants from other states were statistically significant, meaning only that it is unlikely they were explained by chance.

The results confirmed a more detailed expectation that the greatest accentuation of an expansive approach would occur for U.S. claimants, followed by the other three states. That said, there was limited country-specific data for Germany and France. Of the 100 cases, 30 involved a U.S. claimant,[16] 9 a U.K. claimant,[17] 6 a French claimant,[18] and 5 a German claimant.[19] Also, the hypothesis was not supported by the isolated results for Germany. There was no statistically-significant difference between German claimants and claimants from other states; moreover, there was an apparent tendency toward a less expansive approach for German claimants, although this finding was not statistically-significant in that it was accompanied by an unacceptable risk (36%) that the variation was explained by chance.

The strongest finding was that claimants from major Western capital-exporting states who bring claims under a bilateral investment treaty or the Energy Charter Treaty (as opposed to the North American Free Trade Agreement [NAFTA]), and that raise one of four of the coded issues before frequently-appointed arbitrators, are more likely to benefit from an expansive approach. By extension, it can be inferred that a respondent state, although at a disadvantage on such issues relative to investors generally, is more likely to benefit from a restrictive approach where the claimant has the nationality of a state other than a major Western capital-exporter, where the claim is under NAFTA, and where the arbitrators are not frequently-appointed.

The robustness of the findings for the primary grouping (France, Germany, the U.K., and the U.S.) was tested further by analyzing other groupings of claimant nationalities associated with Western capital-exporting interests.[20] Additional statistically-significant evidence was found of an accentuated tendency toward an expansive approach in cases brought by claimants from a G-7 state; a Western European former colonial power; or an OECD member state as of 1990 or 2000. Similar tendencies were observed for other groupings based on UN geographic classifications and World Bank income classifications, but the findings were not statistically-significant.

Conclusions and limitations

The study offers tentative evidence of systemic bias in the resolution of jurisdictional issues in investment treaty arbitration. The tested expectations were that arbitrators would favour legal interpretations that tend to benefit claimants by expanding the authority of tribunals and by allowing more claims to proceed, especially in cases where the claimant is from a Western capital-exporting state. That said, the study also has important limitations. An initial caveat is that the findings do not establish evidence of actual bias on the part of any individual or in any particular case. Even at a systemic level, there is a range of possible explanations for the results, some of which do not entail any inappropriate bias, and further inferences are required to connect the observed tendencies to the underlying rationales for systemic bias. Likewise, there are limitations of the coding process and analytical tools, and of quantitative methods generally, in the examination of adjudicative bias.[21]

The key finding is that the observed tendencies exist in the coded data and are very unlikely to be explained by chance. Notable also is the fact that the observed tendencies reflect variations in the resolution of contested issues arising from ambiguity or silence in investment treaties. Such variations are less likely to be explained by untested factors that may drive case outcomes, such as factual differences among cases or hidden meaning in the text of awards.[22] More broadly, the wider question of possible bias calls for further study, whereas the question of system design should depend ultimately on policy judgments about the system’s structure and processes as evaluated against doctrinal and theoretical principles of adjudication.

Author: Gus Van Harten is Associate Professor at Osgoode Hall Law School of York University, where he teaches Administrative Law, International Investment Law, and Governance of the International Financial System. His research is available openly at www.ssrn.com/author=638855 and at www.iiapp.org.


[1] e.g. Jan Paulsson, Denial of Justice in International Law (Cambridge: Cambridge University Press, 2005) at 265; Charles N. Brower and S.W. Schill “Is Arbitration a Threat or a Boon to the Legitimacy of International Investment Law” (2009) 9 Chicago Journal of International Law 471.

[2] e.g. Olivia Chung “The Lopsided International Investment Law Regime and Its Effect on the Future of Investor-State Arbitration” (2007) 47 Virginia Journal of International Law 953; Ibironke T. Odumosu, “The Antinomies of the (Continued) Relevance of ICSID to the Third World” (2007) 8 San Diego International Law Journal 345.

[3] José A.F. Costa, “Comparing WTO Panelists and ICSID Arbitrators: the Creation of International Legal Fields” (2011) 1(4) Oñati Socio-Legal Series at 3; Catherine A. Rogers, “The Vocation of the International Arbitrator” (2005) 20 American University International Law Review 957 at 1006-7.

[4] Gus Van Harten, “Arbitrator Behaviour in Asymmetrical Adjudication: An Empirical Study of Investment Treaty Arbitration” Osgoode Hall Law Journal (forthcoming). A draft version of the study has been posted at www.ssrn.com/author=638855 along with the dataset and coding notes.

[5] See especially Yves Dezalay and Bryant G. Garth, Dealing in Virtue (Chicago: University of Chicago Press, 1996) at 8-9, 36, 45, 50, 70, 93, 124, and 194.

[6] Drahozal, supra note 5 at 500 and 503; Nudrat B. Majeed, “Investor-State Disputes and International Law: From the Far Side” (2004) 98 ASIL Proceedings 30 at 31.

[7] Walter Mattli, “Private Justice in a Global Economy: From Litigation to Arbitration” (2001) 55 International Organization 919 at 921-2; Rogers, supra note 4 at 968.

[8] e.g. William W. Park, “Arbitrator Integrity: The Transient and the Permanent” (2009) 46 San Diego Law Review 629 at 651-3 and 657-66.

[9] Rau, supra note 19 at 521-2.

[10] Mark A. Hall and Ronald F. Wright, “Systemic Content Analysis of Judicial Opinions” (2008) 95 California Law Review 63.

[11] Fedax NV v. Republic of Venezuela (Award of 11 July 1997), 5 ICSID Rep 186, 37 ILM 1378. The criteria include duration, regularity of profit and return, assumption of risk, substantial commitment, and significance for the host state’s development.

[12] The results summarized here relate to the first two hypotheses of the study. There was insufficient data to test a third hypothesis of systemic bias in favour of France, Germany, the U.K., and the U.S. as respondent states.

[13] To confirm this, a statistical model (a one-sample binomial test) was used by a statistician, Heather Krause, who was employed by the author as a research assistant. The model generated a significance level of p<0.001. Thus, the likelihood that the variation between expansive and restrictive resolutions was explained by chance was very low.

[14] Some arbitrators and commentators present the system in these terms (e.g. Federico Orrego-Vicuña, “Arbitrating Investment Disputes” (2008), posted on the International Council for Commercial Arbitration website).

[15] This was confirmed using a statistical model that controlled for different potential predictors of variances in issue resolutions and that was designed to mitigate the risk of over-statement of the significance or effect of any predictor. These fixed effects were controlled for: claimant’s state of nationality, specific issue among the coded issues, treaty or treaty type, total appointments per arbitrator, and total issue resolutions per case.

[16] As a descriptive finding, in 30 cases brought by a U.S. claimant 47 coded issues were resolved expansively or restrictively (leading to 140 distinct issue resolutions by individual arbitrators in those cases). There were 128 expansive resolutions and 12 restrictive resolutions. The cases were spread among 12 respondent states.

[17] As a descriptive finding, in 9 cases brought by a U.K. claimant 17 coded issues were resolved expansively or restrictively (leading to 49 distinct issue resolutions by individual arbitrators in those cases). There were 42 expansive resolutions and 7 restrictive resolutions. The cases were spread among 6 respondent states.

[18] As a descriptive finding, in 6 cases brought by a French claimant 15 coded issues were resolved expansively or restrictively (leading to 45 distinct issue resolutions by individual arbitrators in those cases). There were 36 expansive resolutions and 9 restrictive resolutions. Notably, 5 of the cases were against Argentina; the remaining case was against the Dominican Republic.

[19] As a descriptive finding, in 5 cases brought by a German claimant 8 coded issues were resolved expansively or restrictively (leading to 24 distinct issue resolutions by individual members of the tribunal in those cases). There were 13 expansive resolutions and 11 restrictive resolutions. The cases were spread among four respondent states (Argentina, Philippines, Russia, and the Ukraine).

[20] These analyses were not based on any detailed country-specific expectations.

[21] Hall and Wright, supra note 6 at 87-8; Klaus Krippendorf, Content Analysis: An Introduction to Its Methodology (Newbury Park: Sage Publications, 1980) at 22.

[22] Brian Leiter, “Rethinking Legal Realism: Toward a Naturalized Jurisprudence” (1997) 76 Texas Law Review 267 at 269; Jonathan P. Kastellec, “The Statistical Analysis of Judicial Decisions and Legal Rules with Classification Trees” (2010) 7 Journal of Empirical Legal Studies 202 at 205-6; Hall and Wright, supra note 6 at 99.