There rarely comes along a case in which a tribunal renders an award in violation of the parties’ agreement on the most fundamental features of their arbitration, such as composition of the tribunal and designation of the arbitral seat. Even more rarely would such an award rendered in violation of the parties’ agreement win recognition or enforcement under the New York Convention or comparable domestic legislation.

Unfortunately both scenarios have recently arisen in the case before the Singapore High Court (the ‘Court’) of Sanum Investments Limited v. ST Group, Limited, and others, [2018] SGHC 141 (18 June 2018).1

The dispute and arbitration before SIAC

The dispute arose out of an agreement (the ‘Master Agreement’) between Sanum, an investment company established in Macau, and several Laotian parties (ST Group Co., Ltd; ST Vegas Co., Ltd; and Sithat Xaysoulivong) to create three joint ventures for the purpose of establishing and operating gaming facilities (i.e. casinos and slot clubs) in Laos.

The Agreement provided that, if a dispute between the Parties could not otherwise be settled, it would be resolved through an arbitration seated in Macau under the auspices of ‘an internationally recognized mediation/arbitration company’.

When a dispute arose, Sanum initiated arbitration against the Laotian entities (referred to - collectively with ST Vegas Enterprise Ltd – as ‘the Laotian Parties’). It did so not before a tribunal seated in Macau, as provided for in the Master Agreement, but before a three-member tribunal seated in Singapore. Sanum brought the arbitration in SIAC, whose default rules mandated a sole arbitrator. According to the Court, the Laotian Parties had promptly objected to the Singapore arbitration on the ground that it was not in accordance with the Master Agreement, and declined to participate in the proceeding. SIAC nevertheless found itself prima facie satisfied that the dispute fell under a separate Participation Agreement, which was concluded by the parties in addition to the Master Agreement, in order to regulate the joint ventures for the slot clubs. This Agreement provided for arbitration of disputes by a three-member tribunal in Singapore under the aegis of SIAC.

In the absence of the Laotian Parties’ participation, SIAC proceeded to appoint all three tribunal members sitting in Singapore. The tribunal eventually rendered an award in favor of Sanum in the amount of US$ 200 million.

The Singapore High Court judgment

The High Court, seized of an action to enforce the award, acknowledged that the arbitration should not in fact have been conducted by a three-member tribunal seated in Singapore, but rather by a single-member tribunal seated in Macau. However, it nevertheless proceeded to enforce the award on the ground that the errors were inconsequential.

As the Court noted, the Participation Agreement did not cover all the slot clubs that were subject to the Master Agreement. It was in particular inapplicable to the so-called Thanaleng Slot Club, which was in fact the entity surrounding which the dispute had arisen. The Court however concluded that the Thanaleng Slot Club was the subject of a separate revenue-sharing agreement and that, since the dispute between the Parties was not governed by the Participation Agreement, it was subject solely to the Master Agreement’s arbitration provision.

The Court acknowledged that the award suffered from serious flaws. First, one named respondent (ST Vegas Enterprise) was not a party to the Master Agreement. The Court concluded it should not have been brought into the arbitration.

More to the point, the High Court found unequivocally that the dispute at issue in the case arose solely out of the Master Agreement and was subject to resolution solely under the Master Agreement’s arbitration provision, which, as noted specifically contemplated arbitration in Macau before a sole arbitrator. It properly found that the Participation Agreement only concerned the two other slot clubs, which ‘are not, and have nothing to do with, the Thanaleng Slot Club’. In sum, according to the Court, ‘the only agreement from which the underlying dispute arose out of is the Master Agreement’.

In reaching that conclusion, the Court found that the tribunal simply assumed, again without evidence, that the arbitration provision set out in the Participation Agreement extended to all disputes arising out of the Master Agreement, including the Thanaleng Slot Club dispute. The High Court squarely disagreed (at [106]):

While the commencement of the arbitration at the SIAC was proper, the tribunal was wrong to have held that the seat was Singapore. The arbitration ought to have been seated in Macau.

Turning to the composition of the tribunal, the High Court found that, since the Participation Agreement was inapplicable, so too was its provision for a three-member tribunal. It held that since the Master Agreement did not specify the number of arbitrators, the tribunal, in accordance with SIAC’s default rule on tribunal composition, should properly have been composed of a single member.

Notwithstanding these errors in seat and panel composition, the High Court proceeded to enforce the award, justifying the result on the ground that ‘the Lao disputants have done little to demonstrate the manner in which these procedural irregularities have affected the arbitral procedure adopted’. The Court found that the breaches – an incorrectly seated arbitration and an incorrect number of arbitrators – were merely procedural in nature and not serious enough to warrant denial of enforcement of the award. The Court conceded that the choice of seat ‘is an important aspect of an arbitration, as the seat indicates the curial court to supervise the conduct of the arbitration’. However, it drew a distinction between set aside and enforcement actions. In its view, the errors that were committed might be treated as sufficiently serious if the Court had before it a set aside rather than an enforcement action. The Court concluded (at [118]):

While there is force in the Lao disputants’ procedural objections, this court is not minded to refuse enforcement of the Award.


The Court’s disposition of the case is deeply disturbing, and for three principal reasons.

First, the Court is mistaken in characterizing the departures from party agreement on seat and composition of the tribunal as merely procedural, and thereby imposing an affirmative burden on the complaining party to show prejudice. Second, the Court failed to appreciate that, given the very nature of the errors committed in this case, affirmatively demonstrating prejudice is simply not practicable. Third, the Court failed to appreciate that denial of the right to seek annulment of an award in a competent court of the arbitral seat is in and of itself prejudicial.

Procedural v. jurisdictional stakes

As noted, the Court justified excusing the departures from party agreement on the ground that the Laotian Parties had not demonstrated that the errors committed had an impact on the proceedings or, by extension, the outcome in the case. In supporting that position, the court characterized the errors as merely ‘procedural’ in nature and therefore ones that, in the absence of a showing of prejudice, did not justify denying recognition or enforcement.

The Court erred in characterizing departures from party agreement on such fundamental matters as the seat and composition of the tribunal as nothing more than procedural concerns. Unlike truly procedural matters, such departures do not pertain simply to the mode or manner in which an arbitral proceeding is conducted. It would be more accurate to describe them as defects of a basic institutional character, but even that is an understatement. The fact is that the parties, in agreeing to arbitrate, had specifically agreed to submit their dispute to a sole arbitrator seated in Macau. This means that only a sole arbitrator seated in Macau would have jurisdiction to hear and decide their dispute. Any tribunal constituted differently or seated elsewhere would lack subject matter jurisdiction, and labelling the errors in this case as if they related to procedural modalities only obscures this important reality.

That the errors in question were jurisdictional is of the utmost importance. In the case of a court judgment, the court’s lack of jurisdiction, whether based on general jurisdictional principles or on an exclusive forum selection clause designating another forum, is fatal to the resulting judgment. A party that can establish lack of jurisdiction should not be, and is not, required to establish harm; no additional prejudice need be shown. This principle, applicable to lack of judicial jurisdiction, is no less applicable to lack of arbitral jurisdiction. A tribunal that exceeds its jurisdiction by adjudicating a claim not properly before it ipso facto produces an unenforceable award.

The difference between procedural and jurisdictional defects in arbitration is fundamental for other reasons as well. The courts’ degree of tolerance of procedural error is understandable. Over the course of an arbitral proceeding, a tribunal is called upon to make myriad procedural determinations. In almost every instance, procedural questions may reasonably be decided in more than one way. How they are decided is a matter of judgment, and arbitrators are selected in large part based on their individual capacity to make judgments of precisely that character throughout the course of the proceedings. Indeed, the exercise of discretion in matters of arbitral procedure goes to the very heart of arbitration. It is for this reason that courts show substantial deference to tribunals’ procedural rulings, overruling them only if their errors reach a certain threshold of seriousness and only if prejudice is shown.

This approach to procedural error is sensible, since the exercise of judicial restraint in review of procedural determinations is critical to arbitration’s very efficacy. To routinely permit judicial ‘second-guessing’ of procedural decisions by an arbitral tribunal would undermine certain of arbitration’s very core purposes, which in addition to favouring economy in time and expense, is designed to ensure a high degree of procedural flexibility in matters of procedure.

Jurisdictional errors, including errors in the composition of the tribunal and the seat of arbitration, are of an entirely different nature. Most obviously, these are not even decisions that arbitral tribunals are authorized to make, unless of course the parties have remained entirely silent on these matters, which is exceptionally rare. These are decisions for the parties to agree upon themselves, as they did in this case, and they are decisions that tribunals must respect. Jurisdictional error implicates the very legitimacy of the proceeding, whereas procedural error does not.

Where jurisdictional error has occurred, there is simply no room for either of the limitations on judicial review noted above: neither a requirement that the violation of party intention was a very serious one, nor a requirement that actual prejudice was suffered. Deviations from the parties’ agreement on composition of the tribunal and seat of arbitration are not matters of degree. They are, by their very nature, i.e. in and of themselves, serious.

Assessments of the counter-factual

It is also entirely unrealistic and impracticable to place on a party whose legitimate expectations as to such fundamental arbitration matters as tribunal composition and arbitral seat have been violated the burden of proving that the errors made in those respects actually affected the proceedings and their outcome.

In the case of most procedural errors, one can intelligently inquire into how differently the case would have unfolded had those errors not occurred. Assessment of the procedurally counter-factual – i.e. giving an account on the assumption of a different fact or set of facts – is of course always to a degree conjectural. But it is seldom impossible. In the case of procedural error, we can ordinarily make something of an educated guess as to whether the errors were outcome-determinative. We try as best we can to engage in some sort of reconstruction of the proceedings on the basis of the procedural issue in dispute having been disposed of differently. What facilitates our ability to do so is that, generally speaking, a procedural determination is a relatively discrete and bounded event.

The errors in this case could hardly be more different. This is especially so when it comes to tribunal composition. The dynamics of arbitration before a sole arbitrator and a three-member tribunal differ enormously and cannot even remotely be quantified. Arbitral decision-making is a human activity; it depends on the personalities of the arbitrators, their approach to arbitration, and countless other personal and professional attributes. The ramifications of panel composition are pervasive. One can simply never know what difference a differently constituted tribunal might make. The consequences are necessarily unknowable. If the only difference were that the cost of a three-member tribunal exceeds the cost of a sole arbitrator, matters could be remedied rather easily. But that of course is by no means the only or the most important difference that a wrongly-numbered tribunal makes.

Prejudice in fact

For the reasons given, I therefore firmly believe that the assimilation of tribunal composition and designation of seat to issues of procedure is deeply misguided, and therefore so too is transposing the requirement of prejudice from errors of procedure to errors in tribunal composition and seat. I also believe that a determination of the counter-factual cannot satisfactorily be made, when it comes to errors of the sort before us, particularly errors in tribunal composition.

On these bases alone, the decision of the High Court in the present case is deeply flawed and, ultimately, unjust. An assessment of the court’s decision could accordingly stop there. But the errors committed in this case – and the error as to the seat in particular – cannot be dismissed as inconsequential. Naturally, the seat determines the nationality of an award for purposes of applying the New York Convention. There is however a great deal more to the seat than that. It determines the law applicable to the meaning and validity of the agreement to arbitrate, and such fundamental matters as non-arbitrability and public policy. That law also governs a vast number of respects in which an arbitral proceeding unfolds. These include such matters as the standards for grant of interim relief, the choice of applicable law, interest and costs, and a host of other highly relevant considerations. Above all, the seat represents the jurisdiction that is competent to set aside an award, thereby bringing into play both the practices applicable to annulment actions in that jurisdiction and the interpretation of the grounds for annulment prevailing there. A party’s right to seek set-aside of an award in the proper forum is fundamental and denial of that right is inherently prejudicial. Disregarding the agreement between the parties as to seat is thus inherently prejudicial.

It is useful to recall here the High Court’s remark that if the action before it had been a set aside action, rather than an enforcement action, it would have treated the errors in the arbitration as more serious, and possibly as grounds that per se warranted annulment. But Singapore was in fact the seat at which the arbitration was held and the award was rendered. As the seat, Singapore has the greatest interest in the legitimacy of the arbitral proceeding and the resulting award, and that interest is just as great in an enforcement action at the seat or in a set aside action at the seat.


For all the reasons stated, the High Court’s casual disregard for the parties’ agreement on the fundamentals of their arbitration is troubling in the extreme. The position taken by the Court amounts to saying that it simply does not matter whether an arbitration was or was not conducted under the arbitration law or by the number of arbitrators specifically agreed upon by the parties. It tells us that, irrespective of the disregard shown for the agreement of the parties on fundamental matters, an award will not be denied recognition or enforcement, even at the seat of arbitration, unless the complaining party carries the burden of proving prejudice. It treats disrespect for consent of the parties on the most basic attributes of their chosen dispute resolution mechanism as of no consequence.

This is not only a dangerous signal for arbitrations to come. But it is inimical to arbitration’s reputation for lawfulness, which these days is all the more important to protect.