[Yash Sinha is an advocate practising before the courts in New Delhi]
Article V of the Convention on the Recognition and Enforcement of Foreign Arbitral Awards (the New York Convention, or “NYC”) confers upon enforcement courts an independent check on foreign awards, one that tests fairness case-by-case. Earlier this year, India’s Supreme Court delivered something else: a fairness check in name, blanket deference in practice.
In Nagaraj V. Mylandla v PI Opportunities Fund-I (“Mylandla”), the Court said it was applying transnational issue estoppel (“TIE”), which bars re-litigation because the issue was already heard and fairly decided. In substance, it applied something else: the primacy principle, which bars re-litigation not on how the issue was heard, but because the seat court is the one that ruled on it. That distinction matters. It leaves Mylandla open to two competing readings, one tied to TIE’s strict conditions, another to broad deference to the seat, all playing out inside a single statutory provision.
Section 48 of the Arbitration & Conciliation Act, 1996 is India’s adaptation of Article V, letting Indian courts refuse enforcement on specified grounds. Mylandla turns on a question the provision leaves open: must the enforcement court examine an objection again once the seat has spoken, or defer to it?
The Dispute
The dispute arose from a Singapore-seated arbitration. A private equity investor had bargained for an exit: an initial public offering (“IPO”), a secondary sale or, failing both, a buyback of shares by the company’s promoters. None happened. The arbitral tribunal awarded damages to the investors and directed them to return their shares once payment was made.
The promoters challenged the award before the Singapore High Court, the seat court, arguing that the share return amounted to an unlawful buyback. Indian law allows companies to repurchase their own shares only in tightly defined situations. The promoters said this was not one of them. The seat court rejected the challenge.
Before the Madras High Court, the enforcement court, the promoters returned with the same objection reframed as a public policy violation under section 48(2)(b): the buyback restrictions, they argued, reflected a policy too fundamental for an award to override. The Supreme Court refused to reopen it, citing TIE [paragraphs 60-83].
The Road to Mylandla
Two decisions before Mylandla already leaned toward TIE. The first emerged from the Delhi High Court in Cruz City 1 Mauritius Holdings v Unitech Limited, which read section 48’s permissive “may” as doing real work: a court can enforce an award even when a section 48 ground for refusal is made out, and that choice is not arbitrary.
It invites the court to weigh how the parties have behaved. And once conduct matters, a party that already raised and lost an objection at the seat cannot be allowed to raise it again simply to hold up an award it dislikes.
The Supreme Court took that reasoning further in Vijay Karia v Prysmian Cavi E Sistemi Srl (“Vijay Karia”), the most significant pre-Mylandla treatment of the question. It split objections into two kinds. A forum-connected objection touches India’s own laws. Jurisdiction and public policy, the very grounds later raised in Mylandla, are examples. A forum-neutral objection, by contrast, touches only the two parties and never India’s own laws, like a complaint about unfair procedure.
In Vijay Karia, this distinction had a consequence. Because the NYC favours enforcement, its grounds for resistance must be read narrowly. For forum-neutral objections, where the enforcement court has no sovereign stake of its own, that narrow reading pointed toward estoppel.
By the time Mylandla arrived, both decisions pointed the same way. Both supported TIE, not deference to the seat. The Supreme Court still had to decide whether to follow that path.
The Test the Court Promised to Apply
Mylandla purported to build on that foundation. It answered the question through TIE, using two authorities. The first was Good Challenger Navegante S.A. v Metal Export Import S.A. (“Good Challenger”), an English Court of Appeal decision on issue estoppel in domestic arbitration. It sets four conditions: (i) the issue must be the same one already decided; (ii) the earlier ruling must have been a real decision on the issue, not a procedural dismissal; (iii) the parties must be identical; and (iv) applying estoppel must not be unfair, meaning the party shut out must have had a real chance to contest the issue.
The second was Republic of India v Deutsche Telekom AG (“Deutsche Telekom”), a Singapore Court of Appeal decision adapting that framework to cross-border enforcement. Deutsche Telekom added a fifth requirement. Before applying TIE, the enforcing court must ask whether the seat’s own law treats the earlier ruling as truly final. InDeutsche Telekom, that meant examining Swiss law before giving its ruling preclusive effect.
TIE is not triggered simply because another court considered a similar issue. It runs through distinct checks, each satisfied before the issue closes. The conduct in Mylandla has to be measured against that checklist.
Did the Court Actually Apply It?
The critical question is whether Mylandla applied the framework it claimed to adopt. On examination, it did not. The Court never conducted a TIE analysis. It asked a different question: had the tribunal and the seat court taken a legally possible view of the buyback issue? Once the answer was yes, the objection was closed [paragraphs 79, 82].
Several TIE features were missing. The Court never examined whether Singapore law treated the earlier ruling as final, as Deutsche Telekom requires. Nor did it conduct a separate fairness inquiry into whether preclusion was justified. One passage exposes the gap most clearly. The Court acknowledged that TIE does not automatically block public policy challenges, because public policy is jurisdiction-specific [paragraphs 76-77].
That mattered here. Before the seat court, the buyback argument appeared as a violation of Indian company law. Before the enforcement court, it arrived wearing a different identity, reframed as a public policy objection under section 48(2)(b). A genuine TIE inquiry would have required the Court to treat that shift seriously and examine the public policy ground afresh. It did not. Instead, the Court treated the seat court’s ruling on the buyback as enough to close the public policy objection itself [paragraph 82]. Whether that ruling was final in the sense TIE requires, or whether preclusion was fair, was never examined. This is hard to explain as TIE. It is easy to explain as primacy.
Two Doctrines, Two Worlds
TIE and primacy are not two names for the same idea. They rest on different premises and produce different outcomes. TIE is party-focused: it asks whether a party has already had its opportunity to contest an issue. Primacy is court-focused: it asks whether the seat court’s ruling, as a ruling, deserves deference regardless of the parties’ conduct. One is about fairness to litigants. The other is about the structural position of the seat.
The distinction is not academic. Major arbitration jurisdictions keep the two apart because each doctrine rests on a different reading of the NYC. TIE courts start from a simple premise: the NYC never made one court superior to another. English and Singaporean courts read that silence as deliberate. The drafters, as Good Challenger notes, could have made seat-court determinations binding on enforcement courts. They did not.
PT First Media TBK v Astro Nusantara International BV follows that logic to its conclusion. The NYC deliberately created two separate remedies. Losing at the seat court is not the same as losing the right to enforce elsewhere. One was never meant to automatically decide the other. Primacy courts read the same treaty and see a hierarchy hiding in plain sight. NYC treats the seat court differently from all other courts. It specifically names the seat court’s annulment as a ground for refusing enforcement.
American cases establish that position. TermoRío v Electranta holds that the NYC gave award annulment power exclusively to the seat court. Karaha Bodas v Pertamina and PLN takes that asymmetry to its conclusion. It holds that allowing the same award to hold different legal statuses in different countries would fracture the very certainty the NYC was designed to create.
Mylandla speaks like a TIE court but reasons like a primacy court, without admitting it has chosen a side. That leaves the judgment open to two competing readings.
One Judgment, Two Readings
On the narrow reading, Mylandla is a TIE case. Future courts must work through Good Challenger’s conditions and Deutsche Telekom’s seat-law inquiry before precluding any issue. On the broader reading, Mylandla is a primacy case. Once the seat court has addressed an issue, enforcement courts defer. It no longer asks whether the TIE requirements have been met. Both can plausibly claim fidelity to the judgment.
That is what makes Mylandla important, and what a future bench will have to resolve. The judgment speaks in the language of one doctrine while reasoning through another, leaving every cross-border award to carry that uncertainty into enforcement. Until a future court resolves that, the opening question remains unanswered: must the enforcement court examine an objection again once the seat has spoken, or defer to it? Both readings of Mylandla are alive, and practitioners must prepare for either.
– Yash Sinha