By Gurman Singh Narula.
The author is a fifth-year student at National Law Institute University (NLIU), Bhopal.
In arbitration law, a recurring difficulty arises when fraud is alleged in the arbitration agreement itself: should such disputes be referred to a tribunal, or filtered out at the threshold? The Supreme Court has addressed this by distinguishing between fraud affecting the underlying contract and fraud directed at the arbitration clause, most recently reaffirmed in Rajia Begum v. Barnali Mukherjee. Yet, while the doctrinal position is settled, the method of its application remains unclear. Courts consistently invoke a “prima facie” standard as a signal of restraint, still its content is undefined, and there is no guidance on the evidentiary threshold, the degree of judicial satisfaction required, or how this inquiry is to remain distinct from a merits-based determination. This indeterminacy produces a structural flaw. A question that ordinarily demands rigorous evidentiary scrutiny is compressed into a summary threshold inquiry without principled limits, leaving outcomes to judicial discretion. In practice, this blurs the line between preliminary review and substantive adjudication, allowing courts to engage in deeper evidentiary assessment than the framework ostensibly permits. The result is a distortion of the arbitral process: the principle of kompetenz-kompetenz is weakened, and jurisdictional authority subtly shifts back to courts under the guise of a limited prima facie examination. To resolve this, this paper seeks to define the content of the prima facie standard by proposing a structured, constraint-based framework that preserves judicial restraint while ensuring principled and consistent adjudication of fraud at the referral stage.
The decision in Rajia Begum v. Barnali Mukherjee exposes a structural gap in arbitration jurisprudence. While courts recognise that disputes involving fraud affecting the arbitration agreement may be non-arbitrable, they have not articulated a clear evidentiary threshold for determining this at the referral stage. The case arose from a disputed Admission Deed, through which one party claimed entry into a partnership and invoked an arbitration clause. At the same time, the other denied the document’s execution, alleging it was forged.
The case is significant for the High Court’s contradictory procedural approach: it refused to appoint an arbitrator under Section 11 due to doubts about the agreement, yet referred the dispute to arbitration under Section 8 using Article 227. The Supreme Court resolved this by holding the dispute non-arbitrable, relying on doubts about the document, prior findings under Section 9, and concurrent lower court rulings, while also finding that the High Court exceeded its jurisdiction.
The judgment does more than resolve a factual dispute; it highlights a deeper systemic concern. By allowing courts to deny arbitration based on an open-ended prima facie assessment of fraud, the law risks inconsistency. Similar cases may yield different outcomes depending on how courts perceive the threshold, gradually weakening the standard over time. While the doctrine rightly recognises that arbitration depends on consent and cannot survive where the agreement is seriously impeached, it offers no clear method for determining when that threshold is met. The result is a framework that is conceptually sound but methodologically underdeveloped.
This article argues that the undefined and indeterminate prima-facie standard is not only a doctrinal gap but a fundamental defect resulting from the tension between the Arbitration Act’s mandate for summary judicial review and the inherently complex evidentiary requirements of adjudicating fraud in arbitral consent.
To remedy this flaw while honouring the principle of kompetenz-kompetenz and minimal judicial intervention, it advances a rigorous three-stage analytical framework centred on a “manifest nullity” threshold. To address this, it advances a three-stage framework centred on a “manifest nullity” threshold, designed to maintain judicial restraint while ensuring that challenges to arbitral consent are evaluated in a structured, consistent, and principled manner.
The evolution of the Court’s approach to fraud and arbitrability reveals a persistent absence of a clear and coherent evidentiary standard governing how such claims should be assessed at the referral stage. While the courts have progressively clarified when fraud may render a dispute non-arbitrable, there is no fixed criterion for evaluating the evidentiary standard at the preliminary stage of a hearing. This article seeks to address this absence by proposing a structured threshold framework.
In A. Ayyasamy v. A. Paramasivam, the Supreme Court drew a foundational distinction: mere allegations of fraud are insufficient to exclude arbitration, but “serious” allegations particularly those that go to the validity of the arbitration agreement itself, may render the dispute non-arbitrable. This marked a shift away from a blanket exclusion of fraud from arbitration toward a more nuanced, consent-based inquiry.
This approach was further refined in Rashid Raza v. Sadaf Akhtar, where the Court formulated a two-pronged test: first, whether the fraud allegation strikes at the arbitration clause itself, and second, whether it implicates broader public law concerns beyond the inter se dispute of the parties. This test sought to operationalise the distinction introduced in Ayyasamy, but still stopped short of specifying how courts should assess such claims at a preliminary stage.
Subsequent decisions, including Avitel Post Studioz Ltd. v. HSBC PI Holdings (Mauritius) Ltd. and Managing Director, Bihar State Food and Civil Supply Corporation Ltd. v. Sanjay Kumar, reaffirmed that allegations of fraud affecting the arbitration agreement raise jurisdictional issues. In such cases, courts are justified in declining reference to arbitration because the very foundation of arbitral authority party consent, is in doubt.
Yet, despite this doctrinal continuity, a critical gap persists. These decisions, including Rajia Begum, identify when arbitration should be refused but do not clarify how courts should evaluate the sufficiency of material at the prima facie stage. The inquiry into consent remains conceptually central but procedurally indeterminate.
The doctrine, when examined closely, reveals a tension that the Court has acknowledged but never satisfactorily resolved. This conceptual tension underscores the need for a more precise threshold, one that can distinguish between mere suspicion and demonstrable invalidity without collapsing into a full merits-based inquiry. The non-arbitrability inquiry is cast as a preliminary jurisdictional exercise, one that must remain summary and avoid devolving into a mini-trial. Yet the question of whether an arbitration agreement is forged cannot be answered without the kind of detailed evidentiary scrutiny that a full trial entails: examination of original documents, expert testimony, and cross-examination.
The “prima facie” standard thus operates as a judicially constructed compromise, but one whose contours remain undefined. This is evident in the Supreme Court’s articulation of the scope of intervention under Section 11. In Duro Felguera v. Gangavaram Port Ltd, the Court confined the inquiry to the existence of an arbitration agreement.
At the same time, the Court’s acknowledgement of flexibility in A. Ayyasamy v. A. Paramasivam, which holds that no rigid rule can be laid down and that each case must turn on its facts, reveals that the prima facie test is not a fixed evidentiary threshold but a context-sensitive standard. This elasticity, while pragmatic, further underscores its lack of doctrinal precision.
When read together, these decisions demonstrate that the jurisprudence on Section 11 does more than limit judicial intervention; it indirectly constructs the prima facie test as a procedural restraint rather than a substantive standard. The cases do not define what degree of satisfaction a court must reach; instead, they define what a court must not do, namely, conduct a detailed evidentiary inquiry. As a result, the prima facie standard derives its meaning negatively, through exclusion, rather than through any positive articulation of evidentiary sufficiency.
The decision in Rajia Begum v. Abdul Rashid illustrates this contradiction in concrete terms. The Court relied on circumstantial indicators, internal inconsistencies in the respondent’s narrative, the prolonged absence of the disputed deed from the documentary record, and contemporaneous banking documents describing her merely as a guarantor to cast a ‘grave cloud of doubt’ over the Admission Deed. These factors are undeniably persuasive. Yet, arriving at a definitive conclusion on authenticity would ordinarily necessitate a full evidentiary process: scrutiny of primary documents, expert analysis (such as handwriting examination), and cross-examination of witnesses. Without these tools, the court’s determination risks resting on an incomplete evidentiary foundation.
This highlights the ambiguity of the ‘prima facie’ standard. Courts invoke it to indicate a limited inquiry, but do not define its content. It is unclear whether it reflects a balance of probabilities, a triable issue, reasonable suspicion, or some arbitration-specific threshold. This uncertainty is not merely semantic, it leads to inconsistent application. Similar cases may be decided differently depending on judicial discretion, with some courts applying the rule of minimal doubt and others requiring stronger proof.
At a deeper level, this exposes a fundamental dilemma within the doctrine itself: a court cannot confidently conclude that an arbitration agreement is forged without undertaking a detailed evidentiary inquiry akin to a trial, yet the statutory framework mandates that the referral stage remain summary and preliminary. The result is a conceptual impasse as the level of scrutiny required for adjudicative accuracy is inherently incompatible with the procedural constraints imposed on the court.
A more serious concern arises from how the Court treats the Section 9 proceedings. The court in the present case held that the High Court’s prima facie finding, questioning the genuineness of the agreement, had attained finality after dismissal of the SLP, and could therefore be relied upon in later Section 8 and 11 proceedings. While this may seem efficient, it raises a doctrinal issue. Section 9 is designed to grant interim relief, with courts undertaking only a summary inquiry and recording tentative findings. In Adhunik Steels Ltd. v. Orissa Manganese and Minerals (P) Ltd., the Supreme Court recognised the interim and provisional nature of such proceedings. Observations made at this stage are not intended to bind subsequent adjudication on merits. Section 9 is meant only for interim relief. The inquiry is summary, the standard is low, and the findings are expressly tentative. Courts routinely clarify that such observations should not affect later proceedings. Allowing a Section 9, especially one made without a full hearing, to block reconsideration at the Section 8 or 11 stage wrongly treats a provisional view as final. It effectively creates a form of estoppel not recognised by the statute. As a result, a party denied relief under Section 9 may also be denied arbitration, leaving them without any effective forum until a civil suit is finally decided, which may take years.
SOLUTION ONE: THE SCHEDULED CATEGORIES APPROACH
The core idea. Rather than asking courts to assess whether a fraud allegation is serious enough on an undefined prima facie standard, which is precisely where the doctrine fails, the law should enumerate, in advance, the specific and closed categories of fraud allegation that are capable of rendering an arbitration agreement non-arbitrable. Outside those specified categories, reference is mandatory. The scheduled approach replaces judicial discretion with a rule: the court does not assess the quality of the fraud allegation; it asks only whether the allegation falls within the schedule. This eliminates the evidentiary standard problem because the court is not conducting an evidentiary inquiry at all; it is performing a classification exercise.
This approach is supported by Vidya Drolia & Ors. v. Durga Trading Corporation, (2021) 2 SCC 1, where the Supreme Court addressed concerns of judicial overreach by replacing open-ended discretion with a structured four-fold test for non-arbitrability. At paragraph 76, the Court held that referral courts must not pre-empt arbitral jurisdiction except where invalidity is manifest, warning that extensive inquiry would undermine kompetenz-kompetenz. The court’s act of classifying four categories of non-arbitrable disputes furthered the approach of minimising judicial discretion and securing the tribunal’s jurisdiction.
Therefore, this approach will solve the problem directly and is rooted in principles established in section 5 and 16 of the act respectively but taking this approach can lead to overarching of domain by the judiciary and this approach will also fail to include new and contemporary fraud patterns making the definitional carving of fraud matters ineffective to solve the problem fully, rather this solution displaces the discretion problem rather than eliminating it.
SOLUTION TWO: THE CALIBRATED THRESHOLD STANDARD (CTS): A PROPOSED TEST
The paper to highlight a key problem: the prima facie standard doesn’t clearly define the kind or amount of evidence required to show fraud. Because of this, courts often apply it inconsistently, leading to the probability of unsupervised judicial discretion. The test of manifest nullity tries to fix this by setting a much higher threshold. However, in practice, it can end up looking very similar to the prima facie test, just with a different label, without truly solving the problem. The mixed approach offers a better solution. Instead of replacing prima facie, it clarifies what the standard actually requires by grounding it in established judicial principles. At the same time, it introduces an upper limit inspired by manifest nullity, ensuring that courts do not overstep. As a result, this approach establishes a standard that is clearer in its requirements, has defined limits, and is consistent with Sections 5 and 16 of the Act.
The Calibrated Threshold Standard (“CTS”) is a two-phase test for pre-reference judicial inquiry where fraud is alleged against an arbitration agreement. It defines the prima facie standard and provides jurisprudential content while imposing a structural ceiling to prevent the inquiry from turning into a trial.
The court must first ensure its inquiry stays within strict limits. It may only consider material placed on record by the parties. It cannot draw inferences from missing documents, assess credibility, identify patterns in circumstantial evidence, or examine authenticity where expert analysis is needed. If resolving the fraud allegation requires any of these steps, the inquiry stops. The matter must be referred to the arbitral tribunal under Section 16 of the Arbitration and Conciliation Act, 1996.
If the inquiry remains within limits, the burden shifts to the party resisting reference. That party must show that the arbitration agreement fails the ‘arguable basis’ threshold, i.e., no reasonable reading of the material supports the existence of an agreement to arbitrate. Mere suspicion, inconsistencies, or late production of documents are insufficient, as they require inference (barred by Phase One). The burden is met only where the material, on its face, makes the agreement implausible, not merely disputed.
The Tie-Breaking Principle
CTS adopts a clear pro-reference rule from Vidya Drolia judgment but applies it in a structured way. It comes into play only after the court has applied both Phase One (the ceiling) and Phase Two (the arguable basis test). If, after the two tests, the court is still whether there is fraud because the material allows more than one reasonable reading but does not clearly show that the agreement is implausible, then the matter must go to arbitration. This kind of ‘doubt’ is specific: it exists where the court cannot decide the issue without drawing inferences or weighing evidence, which it is not allowed to do.
This paper examines the structural flaw at the heart of India’s arbitrability framework: the undefined prima facie standard applied when fraud is alleged against an arbitration agreement. Beginning with the Supreme Court’s decision in Rajia Begum v. Barnali Mukherjee, it traces the doctrinal evolution from Ayyasamy through Vidya Drolia to demonstrate that while courts have progressively refined when fraud may exclude arbitration, the how of that inquiry remains conspicuously unarticulated. The paper exposes the methodological gap between the summary review mandate of Sections 5 and 11 and the evidentiary depth that fraud allegations genuinely demand, and proposes the Calibrated Threshold Standard (CTS) as a principled, structured resolution that preserves kompetenz-kompetenz while ensuring consistency.
The manifest nullity standard, drawn from French arbitration law and Article 1448 of the French Code of Civil Procedure, is theoretically appealing: it counsels courts to intervene only where the invalidity of an arbitration agreement is obvious and indisputable on the face of the record, thereby robustly protecting kompetenz-kompetenz. In principle, it aligns well with the CTS’s ceiling in Phase One. However, transplanting the manifest nullity standard wholesale into Indian law would be constitutionally incongruent. France has not adopted the UNCITRAL Model Law on International Commercial Arbitration, and its arbitration regime reflects a distinct civil-law tradition that operates independently of the Model Law’s framework. India, by contrast, has adopted the Model Law as the structural backbone of the Arbitration and Conciliation Act, 1996. Importing a standard specifically designed for a non-Model Law jurisdiction would create normative dissonance with Sections 5, 11, and 16 of the Act, which reflect the Model Law’s own calibrated approach to judicial restraint and kompetenz-kompetenz.
A more suitable comparative source lies in the approaches of Singapore and the United Kingdom, both of which are Model Law jurisdictions with strongly pro-arbitration cultures that have developed principled, structured frameworks for fraud-related challenges. As demonstrated in Swiss Singapore Overseas Enterprises Pte Ltd v Exim Rajathi India Pvt Ltd and Dongwoo Mann + Hummel Co Ltd v Mann Hummel + GmbH, Singapore’s courts have evolved a three-limb test requiring deliberate concealment, a causative link between the fraud and the award, and an absence of good reason for non-disclosure, all applied against a high evidential threshold. The United Kingdom similarly demands cogent proof of dishonesty under Section 68(2)(g) of the Arbitration Act 1996. Under the Arbitration Act 1996, UK courts enforce arbitral awards via s.66 but robustly intervene where fraud is established. In Contax v KFH [2024], the Commercial Court set aside an enforcement order after finding the entire arbitration agreement, proceedings, and award was fabricated, with sections of the purported award copied verbatim from an unrelated English judgment. Similarly, in Nigeria v P&ID [2023], an $11 billion award was set aside as the arbitration was contaminated by bribery and corruption throughout. English public policy favours enforcement, but fraud must be distinctly pleaded, proved on cogent evidence, and shown to have materially influenced the outcome a deliberately high threshold under ss.67, 68, and 103. India can integrate these features directly into the CTS framework. The “arguable basis” floor in Phase Two mirrors the UK’s triable issue standard, while the pro-reference tie-breaking principle reflects both Singapore’s and the UK’s strong presumption toward arbitration. Adopting this integrated approach would allow India to achieve doctrinal precision without departing from the Model Law architecture that underpins its arbitration statute.
Copyright© 2026 Milon K. Banerji Centre for Arbitration Law instituted at NALSAR University of Law, Hyderabad. All rights reserved.
