Counterclaims by States in Investment Arbitration

It is quite common in investment arbitration for the respondent State to include in its defense to treaty claims one or more criticisms of the investor’s underlying conduct. Such “counterattacks” may include arguments that the investment was illegal from the start, that its operations in due course violated local law, or that the investor breached its direct obligations to the State under a contract. Yet while such arguments feature prominently in State defenses, they are rarely framed as counterclaims seeking affirmative relief. The reason may lie in an instinctive preference by States to pursue any affirmative claims in their own courts. But it may also lie in perceived limits to the jurisdiction of international tribunals to hear State counterclaims. The perception that the institution of investment arbitration is limited to a one-sided presentation of claims, rather than a mutual airing and balancing of claims by both parties, has led to broader criticism of the system.

Two recent ICSID decisions, reaching entirely different conclusions on the issue of jurisdiction over State counterclaims (Spyridon Roussalis v. Romania[1] and Goetz v. Burundi[2]), suggest a need to revisit this issue in a more systematic way. This essay touches briefly on certain jurisprudential and policy factors that may explain the divergent results and frame future cases for further analysis.

First, by way of background, until recently very few cases squarely confronted the issue of jurisdiction over State counterclaims. In some early cases, counterclaim jurisdiction was founded on contract rather than a treaty,[3] and the issue was therefore uncomplicated, since contracts generally allow either party to assert claims for breach. In other cases, tribunals found it unnecessary to address the jurisdictional issue head on.[4] One early tribunal that did consider these issues carefully was Saluka Investments B.V. v. Czech Republic, which found that in principle counterclaims could be heard under a treaty that referred broadly to arbitration of “all disputes … concerning an investment” and incorporated the UNCITRAL Rules, which themselves directly contemplate counterclaims.[5] However, the import of this ruling was significantly limited by two other findings: first, that the tribunal could not hear counterclaims based on breach of a State contract that had its own mandatory dispute resolution clause; and second, that it could only hear counterclaims arising directly as a consequence of the claimant’s having made an investment, not based on general obligations of law applicable to everyone within Czech territory. These two caveats largely swamped the general observation about counterclaims under the treaty.

The Roussalis and Goetz decisions have now revived the issue, by expressly considering the ability of States to pursue counterclaims at ICSID under BITs. The tribunals differed in result, with the majority in Roussalis (Andrea Giardina and Bernard Hanotiau) rejecting jurisdiction over a strong dissent by Michael Reisman, and the unanimous tribunal in Goetz (Gilbert Guillaume, Jean-Denis Bredin and Ahmed El-Kosheri) upholding jurisdiction over counterclaims. The different rulings have engendered debate, both as a strict doctrinal matter involving the sources of consent to State counterclaims in investment arbitration, and as implicating a broader policy debate about the fundamental objectives of investment arbitration.

Because the policy debate lurks not far beneath the doctrinal debate, it is worth bringing it front and center, before examining the more complex issues of consent. First, what are the reasons to allow counterclaims by States? There are several. It may lead to efficiency, to the centralization of inquiry and the avoidance of duplication, all factors that Professor Reisman emphasized in his Roussalis dissent, where he argued that these are “the sorts of transaction costs which counterclaim and set-off procedures work to avoid.” It may avoid inconsistent results in different fora that can engender confusion for the parties and create threats to the legitimacy of the system. It can avoid the sort of impasses that result from anti-suit injunctions and anti-anti-suit injunctions against parallel proceedings, such as have plagued (for example) the many chapters of Chevron v. Ecuador. And it could avoid the irony of a State having moved so far towards acceptance of international arbitration that it embraces it as an alternative to its own national courts, only to be confronted by an investor (who selected arbitration over those local courts for its own claims) insisting on local court exclusivity for the State’s corresponding claims. Professor Reisman in Roussalis criticized the majority for “directing the respondent State to pursue its claims in its own courts where the very investor who had sought a forum outside the state apparatus is now constrained to become the defendant.”

On the other hand, there are important policy reasons also for restricting counterclaims by States in investment arbitration. Since State claims are more likely to arise from concession contracts than BIT obligations (treaties generally do not impose any direct obligations on investors), counterclaims may be seen as an end-run around carefully negotiated contractual dispute resolution clauses. Counterclaims may also embroil tribunals, even more than they are already, in disputes governed solely by local law rather than international law, creating a potential crisis for the legitimacy of tribunals which can claim no greater expertise on such matters than national courts. Finally, extensive use of counterclaims could chill investors from invoking international arbitration against States, and thus potentially defeat the broader goal of BITs to reassure investors by providing an agreed forum for their own claims if and when aggrieved.

The doctrinal debate about sources of consent to State counterclaims has played out, importantly, against this complex policy debate. The first doctrinal question is whether it is sufficient for jurisdiction that both parties have consented to arbitrate under the ICSID Rules.  For this question, it is necessary to unpack the dense text of ICSID Convention Article 46,[6] which couples a mandatory “shall” (“the Tribunal shall … determine any … counterclaims”) with a series of prerequisites (“except as the parties otherwise agree,” “arising directly out of the subject matter of the dispute,” “provided they are within the scope of the consent of the parties” and “provided they … are otherwise within the jurisdiction of the Centre”). The recent decisions have focused mainly on one of these requirements: that the counterclaims “are within the scope of the consent of the parties.” For Professor Reisman (dissenting in Roussalis)[7] and for the Goetz tribunal,[8] the investor’s consent to ICSID was sufficient to imply a consent to counterclaims; there was no need to locate additional or affirmative consent in the underlying BIT. But while this approach may be satisfying from a policy perspective, it arguably is not consistent with Article 46’s own reference to “within the scope of consent” as an extrinsic precondition to the tribunal’s hearing counterclaims. It is worth recalling the bedrock notion that States’ ratification of the Convention does not itself provide their consent to jurisdiction over any particular dispute; rather, consent for any particular claim must be sourced to a writing other than the Convention, such as a treaty, contract, or national legislation. If that is the case for the investor’s claims, why not also for the State’s counterclaims? Stated otherwise, if Article 46 itself provided that consent, then its incorporated requirement of consent (“provided they are within the scope of consent”) would be entirely circular and extraneous.

In this sense, the Roussalis majority’s conclusion that a claimant’s mere filing at ICSID is insufficient in and of itself to create consent to counterclaims is more intellectually robust; the majority reasoned that “the scope of the consent” of the parties referenced in Article 46 must be determined by reference to instruments external to the Convention, such as by the dispute resolution clause contained in the BIT. And certainly, if we examine the different BITs at play in the two recent cases, the differing treaty language provides a more satisfying explanation for the divergent results. The Greece-Romania BIT in Roussalis provided in its Article 9 that “disputes between an investor…and the other Contracting Party concerning an obligation of the latter under this Agreement, in relation to an investment of the former, shall if possible be settled,” or “the investor concerned may submit the dispute” to arbitration (emphasis added).  The majority found this was a narrow offer to arbitrate only investor claims, not a consent to arbitrate State counterclaims— and that it covered only obligations “under this Agreement” (i.e. ones imposed on States), not obligations imposed on investors under local law or contract. By contrast, Article 8(1)(b) of the Belgium-Burundi BIT in Goetz covered disputes concerning the interpretation or application of any investment authorization granted by host State authorities; the tribunal noted that Burundi’s claims about a bank’s alleged noncompliance with its operating certificate fell within this definition.  Article 8(5) of the BIT apparently also referred to national law as well as international law.

For future cases, however, the text of the relevant BIT may not be sufficient to answer all questions regarding jurisdiction over State counterclaims. Contractual dispute resolution clauses may be equally relevant, and could function in one of two ways.  First, a clause could be an independent source of consent to counterclaims notwithstanding a narrow BIT clause:  even if the BIT has a consent clause framed as narrowly as the one in Roussalis, if a concession contract stipulates that claims for breach may be presented to ICSID, that would constitute an independent source of consent, along the lines of the two early Guinea cases addressed above (MINE and Atlantic Triton). But equally, a contract clause could function as a potential “agreement otherwise” within the language of ICSID Convention Article 46, notwithstanding a broad BIT clause; even if a BIT covers “all disputes relating to investments,” which could be seen as reflecting consent to State counterclaims brought under local law, if a concession contract sends contractual claims exclusively to another forum, then arguably this would be a specific “agreement otherwise” precluding that particular class of counterclaims, notwithstanding the broader BIT consent to counterclaims more generally. This was what Saluka held back in 2004, with regard to counterclaims arising directly out of a Share Purchase Agreement. In addition to contractual clauses, future cases also may have to examine issues of possible waiver or possible ad hoc consent to counterclaims. In principle, nothing in Article 46 requires that the investor’s consent to counterclaims appear in the same instrument as the State’s consent to the investor’s claims. After all, consent without privity is a hallmark of ICSID arbitration.

An alternate solution lies, of course, in the development of treaty language expressly to address counterclaim jurisdiction. If future jurisprudence—based on interpretations of Article 46 and varying BIT or contract language—does not reliably permit State counterclaims to be resolved in a single efficient forum with investors’ claims arising from the same subject matter, then States in due course may reform the system to clarify their intent that this occur. This could take the form either of interpretative notes about existing treaty text, or more likely the negotiation of specific provisions about counterclaims in new treaties. Some countries are already beginning to contemplate such provisions, including in proposed model BIT language that would directly impose certain substantive obligations on investors and expressly reference the possibility of State claims against investors for breach of such obligations. In some models currently under review, the investor would have to submit, as a condition precedent to its presenting its own claims, an instrument confirming its advance consent to any counterclaims the State may wish to assert. It remains to be seen, of course, whether such ideas are accepted in any new treaties, and if so whether investors themselves embrace the structure, or alternatively consider it a deterrent to structuring investment through vehicles falling within the scope of the new treaty language. That is a subject for another day.

Author: Jean Kalicki is a partner at Arnold & Porter LLP focusing on international arbitration as both arbitrator and counsel.


[1] ICSID Case No. ARB/06/1 (7 Dec. 2011).

[2] ICSID Case No. ARB/01/2 (21 June 2012).

[3] See, e.g., MINE v. Guinea, ICSID Case No. ARB/84/4 (6 January 1988) (awarding Guinea damages from the investor’s initiation of AAA arbitration in violation of an agreement to resolve disputes before ICSID); Atlantic Triton Co., Ltd. v. Guinea, ICSID Case No. ARB/84/1 (21 April 1986) (rejecting Guinea’s counterclaim on the merits).

[4] See, e.g., Alex Genin et al. v. Estonia, ICSID Case No. ARB/99/2 (25 June 2001) (finding that evidence did not support Estonia’s counterclaim, without expressly addressing the issue of jurisdiction); Gustav F.W. Hamester GmbH & Co. v. Ghana, ICSID Case No. ARB/07/24 (18 June 2010) (finding that Ghana had not pursued its counterclaims after listing them in its initial prayer for relief, and therefore that it was unnecessary to determine whether counterclaims would fall within the scope of consent).

[5] Saluka, Decision on Jurisdiction over the Czech Republic’s Counterclaim (7 May 2004).

[6] “Except as the parties otherwise agree, the Tribunal shall, if requested by a party, determine any incidental or additional claims or counterclaims arising directly out of the subject matter of the dispute provided that they are within the scope of the consent of the parties and are otherwise within the jurisdiction of the Centre.”

[7] See Roussalis dissent (“When the States Parties to a BIT contingently consent … to ICSID jurisdiction, the consent component of Article 46 is ipso facto imported into any ICSID arbitration which an investor then elects to pursue,” so long as the BIT itself does not preclude counterclaims).

[8] See Goetz ¶¶ 278-79 (reasoning that by concluding the BIT, Burundi accepted that disputes could be submitted to ICSID arbitration according to the conditions and procedures in the Convention, including that counterclaims would be evaluated under the conditions in Article 46; by accepting the offer, Goetz in turn consented; therefore jurisdiction exists regardless of whether the BIT contains any provision affirmatively providing jurisdiction over counterclaims).