Enforceability of Arbitration Agreements:

1. Introduction :  Under Indian law, arbitration is a purely consensual, contract-based alternative dispute resolution mechanism. Section 7 of the Arbitration and Conciliation Act, 1996 defines an arbitration agreement as an agreement by parties to submit to arbitration all or certain disputes which have arisen or may arise between them in respect of a defined legal relationship, whether contractual or not. By executing an arbitration agreement, parties exercise their party autonomy to voluntarily waive their fundamental, public law right to have their civil disputes adjudicated by ordinary, state-run civil courts. Because this ouster of civil jurisdiction has grave legal consequences, our courts require strict proof of consensus ad idem—a clear, certain, and concluded meeting of the minds to arbitrate. The Supreme Court of India has consistently held that for an arbitration agreement to be legally enforceable, its terms must demonstrate a present, mandatory, and unconditional obligation to refer disputes to a private tribunal.

2. The Core Attributes of Certainty :  To determine whether an agreement is sufficiently "clear and certain" to be enforced, the Supreme Court has formulated specific structural tests. In the landmark case of Jagdish Chander v. Ramesh Chander 2007 (5) SCC 719, the Court established four key attributes that must be present in any valid arbitration agreement:

1. The agreement must be in writing: Oral understandings to arbitrate are completely invalid under Section 7(3) of the Act.

2. Consensual reference to a private tribunal: The parties must explicitly agree to submit their disputes (present or future) to a private tribunal rather than an ordinary court of law.

3. Impartiality and judicial duty: The private tribunal must be empowered to adjudicate the disputes in a fair and impartial manner, giving both parties a proper opportunity to present their case.

4. Finality of the decision: The parties must have explicitly agreed that the decision rendered by the private tribunal will be final and binding upon them.

This standard is complemented by the six-pronged test adopted in K.K. Modi v. K.N. Modi [1998 (3) SCC 573]  and Bihar State Mineral Development Corpn. v. Encon Builders [2003 (7) SCC 418].  Under this test, a valid clause must contemplate that:

• The decision of the private tribunal is binding.

• The tribunal's jurisdiction is derived from the immediate consent of the parties, a court order, or a statute.

• The tribunal determines the substantive rights of the parties.

• The tribunal owes an equal obligation of fairness and acts in an impartial, judicial manner.

• The agreement is legally enforceable.

• The tribunal decides upon a dispute that is already formulated at the time the reference is made.

If a clause lacks or explicitly contradicts any of these attributes—for instance, by allowing a party to simply file a fresh civil suit if they are dissatisfied with the "arbitrator's" decision—the requirement of certainty is defeated and the clause is rendered unenforceable.

3. The Permissive vs. Mandatory Terminology Dichotomy ("May" vs. "Shall") A critical area where certainty is frequently tested is the choice of language used in the dispute resolution clause. Courts draw a sharp distinction between mandatory terminology ("shall") which creates a binding obligation, and permissive terminology ("may") which indicates a mere option.

• The Wellington Precedent: In Wellington Associates Ltd. v. Kirit Mehta (2000) 4 SCC 272, the Supreme Court examined a clause stating that disputes "may" be referred to arbitration. The Court held that this did not constitute a valid and binding arbitration agreement. The word "may" indicated a tentative choice, meaning that the clause required a subsequent, fresh consent from both parties to actually trigger a reference. In the absence of such mutual consent at the time of the dispute, the court could not compel the parties to arbitrate.

• The Agreement to Agree Rule: This principle was consolidated in Jagdish Chander (2007), where a partnership deed clause stated that disputes "shall be mutually decided by the partners or shall be referred for arbitration if the parties so determine". The Supreme Court ruled that the phrase "if the parties so determine" contemplated a fresh consensus at the time of the dispute. Raveendran, J. held that any clause which requires or contemplates a further consent or consensus before a reference can be made is not an active arbitration agreement; rather, it is merely an "agreement to enter into an arbitration agreement in the future". Such clauses indicate a mere desire or hope to explore arbitration and cannot be legally enforced to compel an unwilling party into arbitral proceedings.

4. The Recent Standard in 2026: The "Can" vs. "Shall" Principle

This strict requirement of certainty was recently reinforced and applied to the permissive word "can" by the Supreme Court of India in the landmark decision of Nagreeka Indcon Products Pvt. Ltd. v. Cargocare Logistics (India) Pvt. Ltd. (2026 INSC 384).

In this case, the dispute resolution clause in a bill of lading provided that any dispute "can be settled by arbitration in India". The appellant argued for a pragmatic, pro-arbitration interpretation, asserting that the clause sufficiently indicated a choice of arbitration. However, writing for the Bench, Sanjay Karol, J. rejected this contention. The Court observed that:

• The word "can", as ordinarily understood, denotes capacity, capability, or factual possibility, rather than a mandatory obligation.

• A clause containing "can" indicates merely a future possibility of referring disputes to arbitration, leaving recourse to ordinary civil courts open.

• To force a party to arbitrate based on such non-binding language would severely compromise party autonomy, which is the absolute sine qua non of arbitral proceedings.

The Supreme Court concluded that permissive terminology—whether "may", "can", or conditional phrases like "if the parties so agree"—lacks the threshold of absolute certainty. For arbitration to be enforced under such a clause, a subsequent, separate agreement is required when both parties actively agree to submit their dispute.

5. Pragmatic Interpretation vs. Fatal Uncertainty : It is important to distinguish between a clause that is vague as to the intention to arbitrate (which is fatal) and a clause that is imperfect as to the machinery of arbitration (which is curable).

• Curing Machinery Defects: Under the doctrine of separability and the principle of minimal judicial intervention, if the intent to refer disputes to arbitration is mandatory and clear, courts will adopt a highly pragmatic and non-pedantic approach to cure procedural gaps. For example, in Enercon (India) Ltd. v. Enercon GmbH (2014) 5 SCC , the arbitration clause was unworkable because it did not provide a mechanism for the two appointed arbitrators to select a third, presiding arbitrator. The Supreme Court held that since the underlying intent to arbitrate was absolutely clear and certain, the court would read into the clause a standard appointment mechanism rather than letting the agreement evaporate into thin air.

• Curing Procedural Ambiguities: Similarly, in Powertech World Wide Ltd. v. Delvin International General Trading LLC (2012) 1 SCC 361, where the clause offered alternative venues ("Arbitrator in India/UAE"), the court held that any initial ambiguity was cured by subsequent correspondence and conduct which conjointly demonstrated a clear, certain, and shared consensus ad idem to proceed with arbitration.

• The Boundary of Judicial Assistance: While courts will actively salvage a poorly drafted clause with a clear mandate, they will never write a mandatory obligation into a clause where the parties used permissive language. If the primary intent to refer disputes to arbitration is conditional, optional, or contingent, the court has no power to cure the defect.

6. Conclusion : Ultimately, an arbitration agreement is a serious commercial contract. For it to be enforceable under Section 8 (referral of parties to arbitration) or Section 11 (judicial appointment of arbitrators), its terms must leave no room for doubt. The terms must clearly and certainly manifest a present, binding commitment to arbitrate. Permissive terms like "may" (Wellington), conditional phrases like "if the parties so determine" (Jagdish Chander), and capability terms like "can" (Nagreeka Indcon) only represent a tentative arrangement to explore arbitration in the future. Without a clear and unconditional consensus ad idem, the courts will protect party autonomy by refusing to compel the parties to arbitrate, leaving them to resolve their disputes in ordinary civil courts.