Whether Hong Kong should permit appeals on questions of law in arbitration is no longer an abstract question. It now arises as part of the broader review of the Arbitration Ordinance initiated following the Chief Executive’s 2025 Policy Address and carried forward by the Working Group on Arbitration Law Reform. That review spans recourse against arbitral awards, tribunal jurisdiction and powers, arbitrability and procedural modernisation. Within that wider exercise, the proposal to make the existing appeal mechanism on questions of law available on an opt-out basis has emerged as one of the most significant, and most contested.
That controversy is unsurprising. Proposals touching arbitral finality invariably attract close scrutiny, particularly where they are framed as altering the default position. But controversy alone cannot determine the policy outcome. The real question is whether, as a matter of legal design and commercial reality, Hong Kong should continue to treat review for material legal error as exceptional, or whether the default should be recalibrated.
It should not.
A confined right of appeal on a question of law is not a retreat from arbitration. Properly designed, it is a limited safeguard that promotes legal certainty, strengthens confidence in arbitral awards and supports the development of Hong Kong commercial law. That is especially so where the parties have chosen Hong Kong law and a Hong Kong seat. In such cases, there is a compelling case that a material misapplication of Hong Kong law should not ordinarily be beyond review.
Comparative experience supports that conclusion. England has retained section 69 of the Arbitration Act 1996. Singapore has also considered the introduction of a similar mechanism, with judicial support for a narrow, leave-controlled right of appeal. A limited appeal mechanism is therefore not inconsistent with the approach of sophisticated, arbitration-friendly jurisdictions.The Existing Framework
This is a narrow mechanism. It is not a rehearing. The court addresses the legal question on the findings of fact in the award, and may confirm, vary, remit or, where appropriate, set aside the award.
The difficulty is that Schedule 2 applies only where the parties opt in, save for limited exceptions. In practice, that has left the mechanism little used. Many parties are unlikely to give focused attention to the opt-in provisions when agreeing an arbitration clause. Hong Kong therefore already recognises the legitimacy of legal-error review, but has made access to it exceptional.
Why the Default Should Change
No serious proposal suggests transforming arbitration into litigation. A right of appeal on law need not involve any rehearing of evidence, reconsideration of factual findings or general review of the merits. It can be confined to pure questions of law, subject to leave, and reserved for decisions that are obviously wrong or, on a question of general importance, at least open to serious doubt.
The English law position remains instructive. Section 69 has long existed without impairing London’s status as a leading arbitral seat. The point is simple: a tightly controlled appeal mechanism does not open the floodgates.
Nor does party autonomy require an opt-in model. Parties choose arbitration for many reasons, including neutrality, expertise, confidentiality, flexibility and enforceability. It does not follow that they have consciously elected to forego all recourse against material legal error. Many arbitration clauses are standard-form provisions adopted without detailed negotiation. Silence is therefore a fragile basis on which to infer an informed choice against appeal.
Indeed, appeals to an assumed market preference for absolute finality are often overstated. In many commercial transactions, parties do not focus closely on the arbitration clause at all, still less on whether a future award should be insulated from a tightly confined right of appeal on law. This is not because arbitration is unimportant, but because the clause is usually agreed at a stage when the parties are focused on the underlying bargain rather than on the architecture of future dispute resolution.
Experience also suggests that user preferences are not uniform. Many parties may indeed prefer complete finality. But others may take a different view, particularly once a dispute has actually arisen. A number of Chinese users, for example, may be more comfortable retaining at least the option of appeal, leaving it to them to decide whether such a right should be exercised in practice. An opt-out model better accommodates that diversity of commercial preference. Parties who truly want complete finality can exclude the right expressly. But where they are silent, the law should not assume that they have knowingly accepted an uncorrectable legal error.
Cost, Delay and Tactical Abuse
The usual practical objection is that a default appeal mechanism would encourage tactical challenges, delay enforcement and increase cost. That concern should not be ignored, but it is manageable.
Hong Kong’s existing Schedule 2 framework already shows how abuse can be controlled. Leave is required. The threshold is high. The question must substantially affect the parties’ rights and must have been one the tribunal was asked to determine. Factual complaints cannot be repackaged as legal questions. The court may also order security and deal firmly with unmeritorious applications in costs.
Those safeguards are substantial. They distinguish a genuine legal-error mechanism from relitigation. Tactical conduct is not unique to appeals on law. Parties already deploy jurisdictional objections, set-aside applications and enforcement resistance for strategic purposes. The answer is not to eliminate every corrective mechanism, but to confine them carefully and supervise them robustly.
There is also a risk of overstating “delay” by treating finality as the only commercially relevant value. Parties care about speed, but they also care about legal reliability. A system that offers no realistic avenue to correct a serious error of Hong Kong law may not always command more confidence merely because it is quicker. Finality remains important. It is not, however, the same thing as infallibility.Public Policy Is No Substitute
Section 81 of the Arbitration Ordinance is directed principally to procedural fairness, jurisdictional defects and public policy. It is not a vehicle for correcting substantive errors of law. Hong Kong courts have consistently made clear that errors of fact or law on the face of an award are not, without more, grounds for setting aside.
Public policy is therefore no substitute for a disciplined appeal mechanism directed to legal correctness. A system that forces parties to recast ordinary legal complaints as due process or public policy challenges is neither principled nor efficient.
Legal Development and Arbitral Quality
The case for reform is not confined to correcting individual awards. It also engages the wider development of the law.
As more substantial commercial disputes move into confidential arbitration, courts have fewer opportunities to articulate and refine commercial principles through reasoned judgments. A limited appeal route helps preserve the relationship between arbitration and the common law by enabling the courts, where necessary, to clarify important points of law.
There is also a quality-control dimension. The possibility of legal scrutiny tends to encourage rigour in decision-making. That does not reflect mistrust of tribunals. It reflects the unremarkable fact that even experienced tribunals may err on questions of law, and that a mature arbitral system should contain a measured means of correction.
Limited appellate scrutiny may also strengthen confidence in arbitral quality. One reason why London arbitration commands continuing market confidence is that the comparatively small body of appealed cases has, in general, demonstrated the resilience rather than the weakness of arbitral decision-making: permission is often refused, and awards are frequently upheld. That is among the most persuasive evidence that a high-quality arbitral system does not fear confined judicial scrutiny, but is instead validated by it.
Institutional Rules and Exclusion
The answer should not be that general finality language in institutional rules automatically displaces the statutory right. Such language ordinarily forms part of a standard procedural framework. It is rarely the product of a specific and informed choice by the parties to waive recourse for material legal error. If a default appeal right is introduced, it should not be neutralised by generic boilerplate.
But the point may go further. It may not be ideal for institutional rules themselves to operate as a direct opt-out mechanism at all. If the object is to respect party autonomy, it is preferable that the choice whether to exclude the appeal right should be made consciously and specifically, rather than through the indirect effect of standard-form rules adopted at the contract stage without focused attention.
A more satisfactory approach may therefore be procedural rather than purely contractual. Shortly after commencement of the arbitration, the administering institution could raise with the parties whether they wish to opt out of the statutory appeal mechanism. That would give them an opportunity to make an informed decision when the dispute is real, the stakes are visible and the consequences of exclusion are capable of meaningful assessment. It would also better reflect commercial reality. Parties negotiating the underlying transaction rarely spend time considering whether, years later, if they lose an arbitration, they should have no possibility of appellate correction on a serious question of law.
Seen in this light, an opt-out model can be defended not only as a legal default, but as a more authentic means of capturing party intention. It preserves complete finality for those who truly want it, while avoiding the fiction that silence in a boilerplate arbitration clause necessarily reflects a deliberate decision to renounce all recourse for legal error.A Measured Reform
The reform required is modest.
First, the right of appeal under sections 5 and 6 of Schedule 2 should apply by default to Hong Kong-seated arbitrations, at least in respect of questions of Hong Kong law, unless expressly excluded.
Secondly, the threshold for leave should remain high.
Thirdly, the appeal must remain confined to law on the findings of fact in the award.
Fourthly, generic finality language in institutional rules should not by itself exclude the statutory right. Exclusion should require a clear and specific party choice. Consideration should also be given to a mechanism by which the administering institution raises the question of opt-out shortly after commencement, so that the parties may decide the matter at a point when their intentions can be formed in a more real and informed way.
That would not weaken Hong Kong’s standing as a leading arbitral seat. On the contrary, it would demonstrate confidence in the balance Hong Kong seeks to strike between finality and legal reliability.Conclusion
A narrow, leave-controlled right of appeal on a question of law would not convert arbitration into litigation. It would preserve finality while recognising that finality does not require the legal consequences of a material error to be left beyond correction. Hong Kong already has the statutory architecture. What is now required is not conceptual innovation, but a more coherent default.
Viewed in the context of Hong Kong’s present arbitration law reform exercise, such a change would be neither radical nor anomalous. It would be a measured refinement of an already sophisticated regime.
The recent controversy surrounding the proposal should not obscure the central point. Finality remains indispensable. But finality should not be mistaken for infallibility. A legal system confident in its arbitral framework should not fear carefully confined judicial scrutiny of material legal error. It should recognise that such scrutiny is part of what makes that framework worthy of confidence.
This article was first published by Hong Kong Lawyer as an online exclusive on 28 August 2026:
Beijing ICP No. 05019364-1 Beijing Public Network Security 110105011258
