Duration of ISDS Proceedings

Written by and

Editors’ Note:  We continue the series of posts (see this morning’s introduction here, and first post on costs here), contributed by individual members of the Academic Forum of the UNCITRAL Working Group III (UNCITRAL WG III) on Investor-State Dispute Settlement Reform, which provide summaries of individual concept papers prepared by various Academic Forum working groups jointly authored by members therein.   These posts will run simultaneously with the ongoing UNCITRAL WG III sessions in New York this week.  This post, authored by Holger Hestermeyer with Anna De Luca, summarizes the concept paper on duration of ISDS proceedings, which was jointly authored by the members of Academic Forum Working Group II (Holger Hestermeyer as chair, Jose Manuel Alvarez Zarate, Crina Baltag, Daniel Behn, Jonathan Bonnitcha, Malcolm Langford, Anna De Luca, Loukas Mistelis, and Gregory Shaffer.)  Clara Lopez Rodriguez and Simon Weber provided assistance.

The length of proceedings, and the resulting impact on costs, has long been a topic of considerable importance for arbitration: a lack of speed is widely perceived by users as one of the downsides of international arbitration. In the current debate about ISDS reform, the length of ISDS proceedings, alongside their costs, has been raised as a concern at the thirty-fourth session of UNCITRAL Working Group III. In the following paragraphs we will try to answer the following questions: when are proceedings excessively long? How long are ISDS proceedings? What is the effect of the various reform proposals on the length of ISDS proceedings?

In carrying out this evaluation the authors faced a number of challenges. Firstly, none of the reform scenarios has been intended to specifically address the duration of ISDS proceedings except for some of the improvements suggested by the scenario “Investment Arbitration (IA) improved”. As a result, the impact on the duration of ISDS proceedings is more of a collateral damage or benefit of the reform proposals under discussion and is, accordingly, not always easy to ascertain. Secondly, the length of investment arbitration is so heavily fact-specific that it seems to defy all attempts at generalisation. Thirdly, short proceedings are not automatically better proceedings. Shortening proceedings can affect dispute resolution in complex ways and involves trade-offs. These have to be made clear. 

When are proceedings excessively long?

The European Court of Human Rights, when analysing whether proceedings take place within a reasonable time under Art. 6 of the ECHR, has held that the reasonableness of the length of proceedings “must be assessed in the light of the circumstances of the case and with reference to the following criteria: the complexity of the case, the conduct of the applicant and of the relevant authorities and what was at stake for the applicant in the dispute” (Comingersoll S.A. v. Portugal [GC], no. 35382/97, § 19) and thus confirmed the fact-specific nature of the determination.

How long are ISDS proceedings?

We will present three datasets on the length of ISDS proceedings: ICSID undertook a review of case duration as part of its rule amendment project, for which it reviewed 63 cases which concluded with an award between 1 January 2015 and 30 June 2017.The review’s result were presented in the following graph:

The second dataset, PluriCourt’s Investment Treaty Arbitration Database, contains 635 cases at the time of data extraction. It yields the following data:





Std. Dev. – Years

Average – All





Average: Decided




















Settled after jurisdiction





Discontinued after jurisdiction










Jurisdiction Loss





Merits Loss





Merits Win






As to annulment, the data set gives the following lengths:






Std. Dev. – Years

Average – All





Average: Decided










Annulment Discontinued after Failure to Pay Fees





Annulment Discontinued










Annulment in Full





Annulment Partial





Annulment Rejected





Langford et al. come to the conclusion that procedural events are the most likely to prolong arbitration: bifurcation, arbitrator challenges and replacement as well as the existence of a dissenting opinion stand out when it comes to explaining longer duration (M. Langford et al., What do we Know?, ISDS Forum).

The third dataset, collected at King’s College London, contains 110 cases. According to it the constitution of an ISDS tribunal took on average of 181 days, 103 days for annulment committees. The written phase of ordinary proceedings without annulment took an average of 407 days. Bifurcated cases took an average of 224 days for the 1st, 379 days for the second and 403 days for the third stage.

What is the effect of the reform proposals on the length of ISDS proceedings?

We will now tackle, one by one, the various reform proposals to gage their possible effect on the length of proceedings.

The first possible reform consists in targeted changes to investment arbitration. Recent ICSID proposals for rule amendments give some guidance as to which changes could be made. Possible changes include speeding up the constitution of arbitral tribunals by triggeringthe default method of appointment automatically; an obligation of conducting proceedings in an expeditious manner; more flexible handling of hearings and witness examinations to avoid the risk of repeating evidentiary exercises in the case of bi- or trifurcation; speeding up the deliberation process and imposing a stricter time limit for delivery of the award; introducing expedited arbitration proceedings with the consent of the parties by a sole arbitrator as proposed for the ICSID Arbitration Rules; simplifying claims consolidation and, where not available, claims coordination; abolishing bifurcation. While these proposals can in some circumstances lead to shorter proceedings, not all of the proposed measures are likely to have a significant impact. Moreover, they involve trade-offs: Speeding up the composition of the tribunal and hearings can limit the right to select arbitrators, and impact the right to be heard, respectively, if the time limits become too strict; similarly strict and too short time limits on writing awards might undermine the quality of awards.

The second proposal is to introduce an appellate mechanism in investment arbitration. The proposal dates back at least to the 2004 proposal for the amendment of the ICSID Arbitration Rules. We do not take a view on the general merits and demerits of such a reform and will focus on its effects on duration. If constituted as the only available mechanism, excluding annulment under the ICSID Convention or the setting aside procedure for non-ICSID awards, an appellate mechanism could save much time. As such, the system could consider the challenge of an arbitral award based on the existence of a procedural irregularity, as set out in Article 52 of the ICSID Convention, as well as for well-defined errors of law and, possibly of fact (within a narrow definition, arguably). Time could also be saved if the appellate body is entrusted with the power to render a new arbitral award where it admits the appeal in part or in full. The resubmission of a case after the annulment proceedings has the potential to add between 5 and 10 years. However, if the proposed appellate mechanism is considered as an addition to the existing remedies against arbitral awards, i.e. setting aside proceedings in the courts at the place of arbitration and annulment under the ICSID Convention, etc., it will necessarily add significant delays in obtaining a final resolution of the dispute.

The third proposal is to establish a Multilateral Investment Court (MIC). It is difficult to predict the procedures of the MIC. In some respects, a MIC model could speed up the proceedings: the constitution of the tribunals would be more expeditious as adjudicators are pre-elected. There would likely be fewer grounds for time-consuming challenges of MIC adjudicators, as they are pre-approved members of a standing-judicial body. However, a MIC is also less flexible to adapt to an increasing caseload, the length of proceedings will depend on the quality of its administration, the appointment process of judges will likely be politicised.

Finally, ISDS could be abolished with interstate arbitration under investment treaties, investment contract-based arbitration and litigation against the host state in domestic courts serving as fallbacks. These options are not mutually exclusive. The analysis of the only three known interstate arbitrations under investment treaties (Chile vs. Peru, Ecuador vs. US, and Italy vs. Cuba), and the US-Canada Softwood Lumber dispute suggest that the resolution of investment disputes through interstate arbitration is roughly comparable in duration to the resolution of those disputes through investor-state arbitration. However, where domestic remedies have to be exhausted, namely where diplomatic protection is exercised, this adds significant delay to the process. Investors and host states can also consent to arbitration of disputes arising under an investment contract between them. Recent ICSID statistics show that 16% of ICSID arbitrations have been based on contractual consent to arbitration. An empirical analysis by Daniel Behn et al. (Why the Delay? Explaining the duration of proceedings in investor-state dispute settlement cases)shows that contract-based ISDS proceedings are slightly shorter in duration. Finally,a comparison of the duration of domestic court proceedings with ISDS raises considerable conceptual and empirical challenges. The duration of domestic court proceedings varies significantly depending on the state in question, the domestic court/tribunal involved, the cause of action and the complexity of the case. Data-sets collated by Bonnitcha, Poulsen and Waibel suggest that investor-state arbitration is, on average, significantly slower than domestic court proceedings in some countries (e.g. Switzerland or Japan), and quicker than domestic court proceedings in others (e.g. India).

Print Friendly, PDF & Email


No tags available

Leave a Comment

Comments for this post are closed