From Scalpel to Sledgehammer: Gayatri Balasamy v. ISG Novasoft and the Perilous Expansion of Judicial Power in Indian Arbitration

[Kanaksha is a student at NALSAR University of Law.]

When the Supreme Court announced in Project Director, NHAI v. Hakeem that Section 34 permitted courts only to set aside an award and never to tinker with its contents, it seemed to cement India’s adhesion to the UNCITRAL Model Law. In Gayatri Balasamy the court reverses course, holding that judges may “modify” an award in four situations—partial illegality, clerical error, oppressive post-award interest, and the amorphous reach of Article 142. The majority dresses the change as pragmatic mercy for parties facing the “nuclear option” of annulment, but the judgment neither acknowledges that Hakeem treated the same argument as textually untenable nor explains why a volte-face was necessary so soon after Parliament’s 2019 amendments that consciously left Section 34 untouched.

To see the practical whiplash, imagine a year‑old construction dispute: Contractor A wins INR 60 crore; Employer B challenges only the award’s timetable analysis, hoping at worst for a retrial. Post‑Gayatri Balasamy, Employer B can instead plead that the timetable portion is “severable” and ask the court simply to lop off INR 15 crore. No retrial, no renewed evidence—just judicial surgery on a tribunal’s anatomy. Nothing in the 2019 amendments signaled this U‑turn, and Hakeem dismissed precisely this “mercy” argument as textually indefensible.

A comparison with the trajectory of England’s Arbitration Act is instructive. There, Parliament itself installed Section 68, empowering courts to step in for “serious irregularity,” but it coupled the power with a one-year limitation period and a requirement of “substantial injustice.” Even then, successful challenges remain vanishingly rare, as the recent Federal Republic of Nigeria v. P&ID decision shows; Mr Justice Knowles treated modification as a remedy of last resort after establishing fraud on an epic scale. India’s judiciary now asserts a comparable power with none of Parliament’s procedural guardrails.

Reading in a Remedy: When Silence is a Deliberate Choice

Chief Justice Khanna’s textual move is elegant: the “greater” power to set aside an award supposedly contains the “lesser” power to prune it. Yet legislative history points the other way. The 1940 Act deliberately used the verb “vary,” and the drafters of the 1996 Act struck it out to mirror the Model Law’s binary. Silence here is not a drafting lacuna; it is a deliberate excision. The Court has therefore supplied statutory words that Parliament purposely withheld, diluting Section 5’s command that courts shall not intervene “except where so provided.” 

Fidelity to Model-Law uniformity carries practical weight. Singapore’s Court of Appeal, in the landmark PT First Media v. Astro dispute, refused to recognize a judicially “corrected” Indonesian award because the seat court’s intervention had muddied the provenance of the instrument. The court stressed that cross-border enforceability depends on recognizing the award as the work of arbitrators, not judges. Gayatri Balasamy exposes Indian awards to the same skepticism: enforcement judges overseas may balk at an instrument that straddles the line between arbitral determination and judicial decree.

Severability: A Scalpel with No Ruler

The majority anoints severability as the “primary technique” for curing partial illegality. In abstract, that seems innocuous; nobody quarrels with excising an utterly discrete head of claim. The difficulty lies in arbitration’s messy realities. Imagine a metro-rail contract where the tribunal awards INR 100 crore on a composite analysis that folds in delay costs, escalation clauses, and counter-claims for defective work. If a court later finds the escalation formula contrary to public policy, it cannot carve the escalation quantum out without re-running technical evidence and recalculating the time-chain—tasks the tribunal alone was equipped to do. Severability, so conceived, drags the court into a granular merits reassessment that Section 34 was explicitly framed to foreclose.

Justice Viswanathan’s dissent anticipates the problem: he points out that “inseparably bound” cannot be a gut‐level intuition. Without objective criteria—such as whether the tribunal recorded separate findings in the operative clause—courts will be tempted to examine witness cross-examination or even re-open expert reports under the guise of checking entanglement. That phenomenon is already visible in Delhi High Court’s recent construction-arbitration docket, where judges spend multiple hearings deciphering technical schedules before deciding whether an award can be sliced. Gayatri Balasamy effectively elevates this intrusive practice to constitutional legitimacy.

Re-pricing Time: The Perils of Judicial Interest Arithmetic

The majority singles out post-award interest as a domain where equity demands flexibility, citing instances of tribunals mechanically uplifting the statutory default of 2% without regard to market forces. The impulse is humane but the implementation disquieting. Interest rates are rarely plug-ins; many tribunals use them to balance asymmetric risks—say, by lowering damages yet compensating delay with a higher rate. Judicial tinkering can therefore upend a carefully calibrated equilibrium.

A concrete illustration comes from infrastructure arbitrations involving state entities. In a 2023 Delhi High Court matter, a public-sector debtor had sat on an award for 3 years, and the tribunal had imposed 15% post-award interest as a behavioural nudge. Under the new regime a debtor can argue that 15% is “oppressive,” invite the judge to shave it to 8%, and thereby shift the economic cost of non-payment back to the creditor. Because the majority fails to define “manifest oppressiveness,” every debtor now has an incentive to plead financial hardship, while creditors must counter-argue with inflation indices, bond yields, and opportunity costs—turning a once summary proceeding into a forensic duel over macro-economic data.

The dissent’s solution—to remit the issue under Section 34(4) for reconsideration by the tribunal—preserves both expertise and party autonomy. International practice backs that view. In England, courts confronting questionable interest computations typically remit under Section 68(3), reserving outright variation for fraud or impossibility. By assuming the role itself, the Indian court invites precisely the merits engagement the Act tried to avoid.

Article 142 as a Hydraulic Pump

The majority reassures that Article 142 will be used sparingly, yet history suggests otherwise. In Perkins Eastman the Court declined to appoint an arbitrator solely under the statute and instead invoked Article 142 to fashion an appointment mechanism of its own design. Trial courts across the country instantly treated that precedent as licence to bypass the statute whenever they spotted a “procedural stalemate.” Once Article 142 hovers over Section 34, litigants will scarcely resist raising equitable pleas, be it for revisiting cost allocation or recalculating foreign-exchange losses. The risk is not theoretical; multiple commercial benches in Mumbai have already entertained oral requests for “minor tweaks” to awards, citing the Supreme Court’s open-ended equity power as an enabling source.

Comparative Perspective: How Other Seats Handle Limited Correction

Pro-arbitration jurisdictions have grappled with the remedial gap between annulment and blind enforcement, but all have relied on legislative precision. England’s Section 68 confines judicial correction to ten enumerated irregularities and imposes a stringent “substantial injustice” test. Singapore’s International Arbitration Act resists modification altogether, confining courts to set-aside or enforcement defenses, yet its Court of Appeal emphasizes that parties retain passive remedies to resist enforcement, maintaining the Model-Law equilibrium. Australia allows remittal under Section 34A of its Model-Law legislation but requires courts to return the issue to the tribunal unless that would be manifestly futile. None of these regimes countenance a blanket judicial power to rewrite substantive relief merely because fairness seems to demand it.

The Dissent’s Blueprint for Coherent Restraint

Justice Viswanathan’s separate opinion, though outvoted, sketches a doctrinally tidy path forward. He recommends strict adherence to the two existing safety valves: Section 33 for slip corrections and Section 34(4) for targeted remand, both subject to the tribunal’s consent and within a capped timeframe. That model preserves international uniformity and places the corrective task back in arbitral hands, where the evidentiary context resides. It also dovetails with the Viswanathan Committee’s draft amendment, which proposes a narrowly framed statutory power of modification “in exceptional circumstances,” subject to defined thresholds and a non-extendable ninety-day limitation. The draft awaits Parliamentary deliberation; adopting it would reconcile equitable pragmatism with statutory fidelity.

Toward Predictable Reform

If the business community genuinely needs a mid-level remedy between annulment and blind enforcement, the remedy must be etched by Parliament, not improvised by the bench. The Legislature can borrow from Section 68’s architecture: require a written application, impose a one-year outer limit, confine grounds to fraud, bias, or patent mathematical error, and mandate that courts first consider remittal. Absent such text-based constraints, Gayatri Balasamy risks turning Section 34 into an elastic clause whose scope expands with every imaginative pleading.

Conclusion: A Step Toward Fairness, a Slide from Certainty

Gayatri Balasamy seeks to avoid the blunt injustice of all-or-nothing review, but its doctrinal scaffolding is built on inference rather than enactment. The decision trades statutory clarity for equitable reach, relying on undefined thresholds for severability, “oppressive” interest, and constitutional discretion. Comparative seats show that limited modification can coexist with finality, yet they achieve that coexistence by legislative detail, not judicial aspiration. Unless a clarifying amendment soon follows—or lower courts impose self-denying discipline—the new power to “edit” awards may reduce one inequity only to replace it with a larger one: the return of unpredictable, merits-sensitive litigation that Indian arbitration law spent a generation trying to escape.