Three times in six years Parliament rewrote parts of the Arbitration and Conciliation Act 1996 (Act 26 of 1996), and a fourth rewrite has been circulated in draft for consultation. The Arbitration Act amendments of 2015, 2019 and 2021 are usually described as one continuous reform movement towards speed, neutrality and institutional arbitration. Read as a trail rather than as a slogan, they tell a less tidy story. The amendment of 2015 changed the daily behaviour of courts and tribunals and largely worked. The amendment of 2019 built an institutional architecture, a council, a grading regime and an appointment scheme, and then left the centre of that architecture uncommenced for the whole of its existence. The amendment of 2021 corrected one excess of 2019 by deleting a qualification schedule that had excluded foreign practitioners, while restoring, for a defined class of cases, an unconditional stay on enforcement that the 2015 reform had abolished. Between the second and third amendments sits a judgment that struck down a provision of the 2019 Act as manifestly arbitrary, an event without parallel elsewhere in this statute’s history.

This article follows the trail provision by provision. For each amendment it separates four questions that secondary writing habitually merges: what the amendment enacted, what was brought into force, what the courts did with it, and what survives. The separation matters because the sharpest fact in this territory is a commencement fact. A reader who consults the text of the 2019 amendment will find an Arbitration Council of India designed in detail, with a chairperson, members, functions and a grading power. A reader who consults the commencement notifications will find that the Part creating the Council was never notified, so the Council has never existed, has never graded an institution and has never accredited an arbitrator. The statute book and the working law are, on this point, two different documents, and the trail is the record of how they came apart.

The Arbitration Act Amendments at a Glance: Three Amendments, Three Fates

The amendment of 2015 was the Arbitration and Conciliation (Amendment) Act 2015 (Act 3 of 2016), deemed to have come into force on 23 October 2015, the date of the ordinance it replaced. It was built on the 246th Report of the Law Commission of India, submitted in August 2014 under the chairmanship of Justice A. P. Shah, and it was a discipline statute: it narrowed what a court may examine when appointing an arbitrator, imposed independence and disclosure duties backed by two schedules of disqualifying relationships, put a twelve-month clock on the making of awards, confined the public policy ground of challenge, and ended the practice under which the mere filing of a challenge suspended enforcement of an award. Almost every provision of that amendment entered into force on the single appointed day, and almost every provision survives, although two of its mechanisms, the timeline and the appointment gate, were recast in 2019.

The amendment of 2019 was the Arbitration and Conciliation (Amendment) Act 2019 (Act 33 of 2019), which received assent on 9 August 2019. It was built on the report of the High Level Committee chaired by Justice B. N. Srikrishna, a former judge of the Supreme Court, submitted on 30 July 2017. Its ambition was institutional rather than disciplinary: arbitration in India was to migrate from courts and ad hoc appointments to graded arbitral institutions supervised by a statutory council. Its fate was bifurcated. One group of its provisions, amending Sections 17, 23, 29A, 34, 37, 45 and 50 of the principal Act and inserting Sections 42A, 42B and 87, was brought into force on 30 August 2019 by notification S.O. 3154(E). The other group, containing the definition provisions, the rewritten appointment scheme in Section 11, the whole of Part IA creating the Arbitration Council of India, and the Eighth Schedule of arbitrator qualifications, was never notified at all. Within three months of the partial commencement, the Supreme Court struck down Section 87, the provision that had confined the 2015 amendment to arbitrations commenced on or after 23 October 2015.

The amendment of 2021 was the Arbitration and Conciliation (Amendment) Act 2021 (Act 3 of 2021). It replaced an ordinance promulgated on 4 November 2020, was passed by the Lok Sabha on 12 February 2021 and by the Rajya Sabha on 10 March 2021, received assent in March 2021, and is deemed to have come into force on 4 November 2020. It did two things. It inserted a further proviso into Section 36(3) directing an unconditional stay of an award where a prima facie case of fraud or corruption is made out, and it deemed that proviso to have been inserted with effect from 23 October 2015, the commencement date of the 2015 amendment, with an explanation applying it to court cases regardless of when the arbitral or court proceedings began. And it omitted the Eighth Schedule, substituting Section 43J so that qualifications, experience and norms for accreditation of arbitrators would be specified by regulations rather than by the statute. It is the only one of the three amendments whose operative reach runs backwards across the entire life of the reformed statute.

Which amendments to the Arbitration Act are in force in full?

Only the amendment of 2015 operates in full, as recast by the notified provisions of 2019. The amendment of 2019 operates only in the parts notified on 30 August 2019; its Council, its appointment scheme and its qualification schedule never commenced. The amendment of 2021 operates in full and reaches back to 23 October 2015 for its stay provision.

That summary conceals an asymmetry worth pausing on. Commencement in Indian legislation is an executive act performed after enactment, and a provision that Parliament has passed but the executive has not notified is law in the sense that it sits in the Gazette, and not law in the sense that any court, party or arbitrator can act on it. The series of which this article forms part has traced that gap in labour codes, in tribunal statutes and in welfare legislation. The arbitration trail is the most striking instance in the commercial field, because the unnotified provisions were not a schedule or a penalty clause but the institutional core of the reform: the body that was to grade institutions, accredit arbitrators and receive designations from the Supreme Court and the High Courts. The appointment system that actually operates in India, in which the courts appoint arbitrators under Section 11 in the form the 2015 amendment gave it, is precisely the system the 2019 amendment was enacted to retire.

How the trail was made: ordinances, expert reports and commencement

The method of the trail is part of its substance. Each of the three amendments was preceded by an expert document: the Law Commission’s 246th Report of August 2014 for the amendment of 2015, the Srikrishna Committee’s report of July 2017 for the amendment of 2019, and, in place of a report, a stated enforcement concern about fraud-procured awards for the amendment of 2021, with the Viswanathan Committee of 2024 completing the pattern for the draft that followed. The sequence gave the statute an unusual quality for Indian commercial legislation: its texts were drafted against published diagnoses, so the distance between the diagnosis and the enacted remedy can be measured in each case, and the shortfall, where there is one, is visible rather than a matter of inference.

Two of the three amendments travelled by ordinance. The 2015 amendment began as an ordinance of 23 October 2015, and the date of that ordinance became, by the device of deemed commencement, the most consequential date in the field, the line on either side of which applicability was fought for six years. The 2021 amendment began as an ordinance of 4 November 2020, deemed into force from that date when the Act followed. Ordinance-making is a constitutionally sanctioned route with a democratic cost that this series has recorded in other fields: the text is settled before either House debates it, and the debate, when it comes, is conducted under the pressure of a law already operating. In arbitration the cost was compounded, because the ordinance dates were not mere administrative conveniences; they were chosen, or at least retained, as the pivot of retrospectivity, and they entered the statute’s litigation history as substantive lines.

The expert documents themselves were produced by distinguishable methods, and the differences left marks on the texts. The Law Commission’s 246th Report was the product of a standing law reform body working from a reference on the statute’s working, and its proposals arrived with draft clauses and a comparative survey of foreign seats. The Srikrishna Committee was an ad hoc body of practitioners, judges, officials and institutional representatives assembled for the arbitration question alone, and its report ranged beyond amendment into institution-building, model contract clauses for government, and the creation of a specialist cadre, so that the 2019 amendment enacted the statutory fraction of a wider programme whose non-statutory fractions, a model government arbitration policy among them, were pursued administratively or not at all. The Viswanathan Committee of 2023 to 2024 worked with the narrowest brief and the most technical method, examining the Act section by section against case law accumulated since 2015, and its draft Bill reads accordingly, as a repair document rather than a manifesto. The pattern across the three exercises is a declining ratio of architecture to repair, which tracks the trail’s own lesson: the first report could propose a system because the system’s failures were systemic, while the third confined itself to defects that a decade of construction had located with precision.

The third feature of the method is the two-step structure of Indian commencement, under which enactment and operation are separate events connected by a notification in the Gazette. The trail contains the full spectrum of what that structure permits: total commencement on a deemed date in 2015, partial commencement with the centre withheld in 2019, and commencement with express backward reach in 2021. A reader of the bare amended Act sees none of this. The Act as printed carries Part IA without any mark that the Part is inoperative, carries the substituted Section 43J without any mark that the framework it serves is dormant, and no longer carries the Eighth Schedule at all, so the printed text understates the 2019 design and overstates its own operation at the same time. The trail method of this article, separating enactment from notification from survival, is the corrective the record requires.

The three fates correspond to three theories of what was wrong. The 2015 amendment proceeded on a behavioural theory: the actors in the system, courts, parties and arbitrators, responded to incentives the statute set, and changing the incentives, a narrower gate, a clock, a costs rule, a stay that must be bought, would change the behaviour. The 2019 amendment proceeded on a structural theory: behaviour follows structure, and a system of ad hoc appointments supervised by generalist courts would reproduce its defects under any incentive scheme, so the structure itself, institutions, grading, accreditation, had to be built. The 2021 amendment proceeded on neither theory in general form; it was an exception carved for a class of cases the system’s defenders and critics agreed were pathological, tainted awards, and its generality is confined to its retrospectivity. The verdicts the trail supports track the theories. The behavioural amendment worked because behaviour is cheap to change: no institution had to be created for Section 36 to be applied by every executing court in the country on the morning it commenced. The structural amendment failed at the point structural reforms fail, the creation of the institution, which required a second decision, sustained executive attention, and the displacement of interests that the first decision had only announced. The exception amendment sits between the two, operative from its deemed date and dependent for its reputation on the discipline of its administration.

The statute the amendments inherited

The parent article in this series owns the Arbitration and Conciliation Act 1996 as enacted, and its account is not repeated here. What the trail requires is only the condition of the statute at the moment the first amendment arrived, because the amendments of 2015 are unintelligible except as answers to named failures.

Three failures dominated. The first was the breadth of the public policy ground for setting aside an award. In Oil and Natural Gas Corporation Ltd v Saw Pipes Ltd (2003) 5 SCC 705, the Supreme Court read the public policy ground in Section 34 to include an award that was patently illegal, and the phrase travelled far beyond the case, converting a narrow review ground into a rehearing on points of law for domestic awards. The second failure was enforcement. The unamended Section 36 was read, in National Aluminium Co Ltd v Pressteel and Fabrications (P) Ltd (2004) 1 SCC 1, to suspend enforcement automatically once a Section 34 application was filed, so an award holder who had won after years of arbitration returned to court and waited again while the award lay frozen, with no deposit and no security exacted as the price of the freeze. The third failure was time. The 1996 Act, faithful to the UNCITRAL Model Law on which it was based, imposed no deadline for making an award, and Indian arbitrations, predominantly ad hoc and frequently conducted by retired judges sitting in short sessions spaced weeks apart, acquired durations comparable to the suits they were meant to displace.

Appointment practice compounded each failure. Section 11 committed appointments, where a party defaulted, to the Chief Justice or a designate, and the resulting applications became a litigation tier of their own, argued on the merits of the underlying dispute, on the validity of the agreement and on the arbitrability of claims, all before the tribunal existed. The doctrinal warrant for that breadth was SBP and Co v Patel Engineering Ltd (2005) 8 SCC 618, where a seven-judge bench characterised the Chief Justice’s appointment function as judicial rather than administrative, with the consequence that the appointing authority was understood to decide, finally, preliminary questions of the agreement’s existence, validity and the claim’s survival. Appointment litigation under that dispensation could run for years and generate its own appeals, and the Law Commission’s 246th Report treated the decision’s practical legacy as one of the principal defects to be legislated away, recommending that the appointing court be confined to the existence of the agreement and that the finality of its order be stated in the Act. The 2015 amendment enacted both recommendations in Sections 11(6A) and 11(7), and in doing so reversed, by statute, the working effect of a Constitution Bench decision, an early instance in the trail of the legislature and the Court correcting each other’s arbitration work in turn, each time within the forms the Constitution allows. Arbitrators drawn from the engineering services of government undertakings sat on disputes to which their own employers were party, and the Act supplied no disclosure regime and no disqualification for interest beyond the general duty in the original Section 12. The Law Commission’s 246th Report catalogued these defects and proposed the architecture that the 2015 amendment adopted with modifications: a prima facie appointment gate, schedules of ineligibility, a timeline for awards, a costs regime that followed the event, and a Section 34 ground confined by explanation. It is against that catalogue that the first amendment must be measured, because the amendment’s authors measured themselves against it.

One boundary of the trail should be marked at the outset, because it disciplines everything that follows. The amendments are, in substance, a Part I story. Part II of the 1996 Act, which enforces foreign awards under the New York and Geneva Conventions, was touched by the trail only at the margins, in the referral and appeal provisions amended among the notified clauses of 2019, and the foreign award enforcement regime that Indian courts administered at the start of the period is the regime they administer at its end, developed by case law rather than by amendment. The explanation is partly textual, in that the Convention framework constrains the room for domestic redesign, and partly a matter of where the pain was felt: the system’s crises of delay, interference and frozen enforcement were crises of domestic and India-seated arbitration, and each amendment aimed at the seat’s own machinery. A reader tracing the trail for its effect on a foreign award’s enforcement in India will find the record thin by design, and the thinness is itself a finding about the legislature’s diagnosis.

The amendment of 2015: discipline imposed on a drifting statute

The 2015 amendment began as an ordinance promulgated on 23 October 2015 and was enacted as Act 3 of 2016, with its commencement deemed to run from the ordinance date. The ordinance route was a habit of that legislative season in commercial law, and the same day produced the companion ordinance from which the commercial courts legislation followed, a pairing examined in the article on the Commercial Courts Act 2015. What distinguishes the arbitration amendment is the completeness of its commencement: unlike its successor of 2019, it took effect as a whole, so the reformed statute that practitioners met in 2015 was the statute Parliament had passed, and the questions that followed were questions of reach and interpretation rather than of existence.

The appointment gate was rebuilt first. The substituted Section 11 confined the Supreme Court and the High Courts, when approached on a party’s default, to a single inquiry. Section 11(6A) directed that the court, or a person or institution designated by it, confine itself to the examination of the existence of an arbitration agreement, and Section 11(7) made a decision under the section final. Gone, in the text at least, was the roving inquiry into the merits, the limitation position and the arbitrability of the claim that had made appointment applications a preliminary trial. The Supreme Court spent the following years policing the boundary of that confinement, and its later reading, that the court performs a prima facie review and leaves contested questions to the tribunal, is the gloss under which the section works. The confinement mattered to the trail because the 2019 amendment proposed to remove the court from the appointment function altogether, and the comparison between a confined court and a displaced court is one of the two or three genuine design choices the trail contains.

The courts spent the decade after 2015 defining the width of the gate that Section 11(6A) had built. In Vidya Drolia v Durga Trading Corporation (2021) 2 SCC 1, a three-judge bench articulated the prima facie test that governs both referral under Section 8 and appointment under Section 11: the court examines whether an arbitration agreement exists and whether the dispute is arbitrable on a first-sight basis, resolving contested questions of fact and of mixed law in favour of referral and leaving them to the tribunal, and intervening only where the claim is manifestly dead, time-barred beyond argument, or within a category the law reserves to public fora. The doctrine of competence-competence in Section 16 does the rest of the work, since a tribunal trusted to rule on its own jurisdiction is the natural repository of every question the appointing court is told to leave alone. The gate was tested from an unexpected direction by the stamping cases. In N. N. Global Mercantile (P) Ltd v Indo Unique Flame Ltd, a Constitution Bench held in 2023 that an arbitration agreement in an instrument chargeable to stamp duty and unstamped is not enforceable until the defect is cured, a holding that threatened to add a revenue inquiry to the appointment stage and to stall appointments across the commercial courts. A seven-judge bench, in the reference reported as In Re: Interplay Between Arbitration Agreements under the Arbitration and Conciliation Act 1996 and the Indian Stamp Act 1899, decided in December 2023, held that inadequacy of stamping is a curable defect that does not render the arbitration agreement void or unenforceable at the referral or appointment stage, and that the examining court leaves stamping objections to the tribunal and the revenue authorities. The two decisions, delivered within one year, restored the 2015 gate to its intended width: the appointing court asks whether an agreement exists, and almost nothing else. The episode belongs to the trail because it shows the amendment’s discipline being maintained against a doctrine that would have dissolved it, by the same institution, the Supreme Court, whose construction of Section 26 the legislature had tried and failed to displace.

Independence was converted from an aspiration into a code. Section 12 was recast to require a prospective arbitrator to disclose in writing any circumstances likely to give rise to justifiable doubts as to independence or impartiality, and the Fifth Schedule set out the relationships that guide that inquiry, drawn from the International Bar Association’s guidelines on conflicts of interest. Section 12(5) went further: a person whose relationship with a party, counsel or the subject matter fell within the Seventh Schedule was ineligible to act, regardless of disclosure, and the ineligibility could be waived only by an express agreement in writing made after the disputes had arisen. The schedules did slow work on the public sector’s appointment culture, under which a government undertaking named its own serving or retired officers to tribunals in its own disputes; a serving employee of a party fell within the Seventh Schedule, and the nomination of a panel controlled by one party became litigable in a way it had not been before. The judicial terminus of that line of the trail, the Constitution Bench decision on unilateral appointment clauses, is taken up at the end of this article.

Interim relief was reallocated between court and tribunal. The amended Section 9(3) provided that once the arbitral tribunal has been constituted, the court shall not entertain an application for interim measures unless it finds that circumstances exist which may not render the remedy under Section 17 efficacious. In the other direction, Section 17(2) gave an order of the tribunal the status of an order of the court for enforcement purposes, so the tribunal’s interim directions ceased to depend on a party’s voluntary compliance or a fresh court proceeding. Read together, the two changes stated a preference: interim protection belongs to the tribunal once the tribunal exists, and the court’s role is front-loaded to the period before constitution and to cases where the tribunal’s remedy would not bite. Section 9(2) added a discipline of its own, requiring arbitral proceedings to commence within ninety days of an interim order made before constitution, a provision that converted the pre-arbitration injunction from a parking place into a trigger.

What did the 2015 amendment do to interim relief under Sections 9 and 17?

It reversed the default. Before the amendment, parties returned to court for interim protection throughout the arbitration. After it, Section 9(3) bars the court from entertaining an interim application once the tribunal is constituted, unless the Section 17 remedy before the tribunal would not be efficacious, and Section 17(2) makes tribunal orders enforceable as orders of the court.

Time was made a statutory obligation. Section 29A required the award in a domestic arbitration to be made within twelve months of the tribunal entering upon the reference, extendable by six months with the consent of the parties, and thereafter only by the court, which could reduce the arbitrator’s fees where the delay was attributable to the tribunal and could substitute arbitrators where it extended the period. Section 29B offered a fast track under which parties could agree, before or at the constitution of the tribunal, to a decision on documents alone within six months. The timeline provisions were the most argued-over part of the amendment in practice, and the least stable: the starting point of the clock was moved by the 2019 amendment, as the account of that amendment records, and the outer architecture of extensions generated its own litigation. What the 2015 text established, and what survived every later recasting, was the principle that the duration of an Indian arbitration is a matter of statutory discipline rather than of diary convenience.

The timeline generated an operational jurisprudence of its own. Parties discovered that the twelve-month period could be consumed by the preliminaries of constitution, service and pleadings, which is the defect the 2019 recast addressed by moving the trigger to the completion of pleadings. Courts were asked whether an extension application filed after the period had expired could be entertained at all, and High Courts divided before the Supreme Court held, in Rohan Builders (India) (P) Ltd v Berger Paints India Ltd in 2024, that the court’s power to extend subsists and that an application made after expiry is maintainable, since the alternative would terminate mandates for a default the section exists to manage rather than to punish. The fee-reduction power in the section, directed at tribunals responsible for delay, has been sparingly used, and the substitution power more sparingly still, which practitioners attribute to the reluctance of courts to record findings against arbitrators who are frequently retired members of their own judiciary. The mandate-termination consequence, under which the arbitrators’ mandate ends if the period is not extended, remained the provision’s real sanction, and its shadow disciplined tribunal scheduling in cases where no extension was ever sought.

The fast track of Section 29B had a quieter career, and its quietness is informative. The provision required the parties’ agreement before or at the tribunal’s constitution, a decision on the basis of written pleadings and documents, no oral hearing unless all parties requested one or the tribunal considered it necessary, and an award within six months of the tribunal entering upon the reference. Uptake remained confined to a narrow class of disputes, principally document-heavy claims of moderate value where the evidence was contractual and the credibility of witnesses was not in issue. Practitioners attribute the limited use to the election’s timing and irreversibility: parties must commit to a documents-only process before the dispute’s shape is known, and a party that later discovers a need for cross-examination has bargained the remedy away. Institutional rules filled part of the gap with expedited procedures that a party may invoke after the dispute has arisen and that the institution grants on a view of the case, a design that allocates the election to a neutral decision-maker rather than to mutual consent at the contract stage. The contrast between the statutory fast track and the institutional expedited procedure is a small instance of the trail’s large pattern: where the statute fixed a mechanism in advance, practice preferred the flexible instrument, and the reforms that endured are those, like the stay and the clock, that set consequences rather than choreographies.

Money was disciplined alongside time. A new Section 31A enacted a costs regime under which the unsuccessful party is ordinarily ordered to pay the successful party’s costs, departing from the older practice under which costs were nominal or unaddressed and each side bore the expense of a process whose length the losing side had often engineered. The Fourth Schedule supplied a model fee structure for arbitrators in court-appointed cases, scaled to the sum in dispute, in an effort to displace the practice of per-hearing fees that rewarded adjournment. The fee schedule had a partial life only, since parties and institutions remained free to agree their own terms, but the costs rule changed settlement arithmetic across the board, and it is one of the few 2015 changes that no later amendment touched. The schedule’s operation in practice divided along the line between court-appointed and party-agreed arbitration. In court appointments, High Courts framed the Fourth Schedule model into their own fee directions and applied its slabs, which are proportioned to the sum in dispute with a ceiling for very large claims, and the per-hearing billing that had made adjournment profitable declined in that segment. In administered arbitration, institutional fee tables governed by the parties’ adoption of institutional rules, and in ad hoc appointments by agreement the schedule operated as a negotiating reference rather than a rule. A friction persisted at the top of the market, where arbitrators of international standing declined appointments priced by a domestic model schedule in disputes seated in India under Indian court appointment, a friction the institutions cite when they argue for appointment routes that bypass the schedule altogether. The institutional redesign of 2019 would have recast the fee question along with the appointment question, and since that redesign never commenced, the model schedule continues to stand in the Act and to govern court appointments.

How did the 2015 amendment change who bears the costs of an arbitration?

Section 31A made costs follow the event as the general rule, so the unsuccessful party is ordinarily ordered to pay the successful party’s costs, and it listed party conduct, including frivolous claims and dilatory behaviour, among the circumstances a tribunal weighs. Before 2015 the Act left costs largely to unguided discretion, and awards frequently said nothing about them.

The challenge ground was confined by drafting technique as much as by declaration. For Section 34, the amendment added an explanation to the public policy clause declaring that an award is in conflict with the public policy of India only if its making was induced or effected by fraud or corruption, or if it contravenes the fundamental policy of Indian law, or if it conflicts with the basic notions of morality or justice, and it barred review on the merits under the cover of that inquiry. A new Section 34(2A) then reintroduced a form of the patent illegality ground in express terms, but confined it to arbitrations other than international commercial arbitration, and declared that an award shall not be set aside merely on the ground of an erroneous application of the law or by reappreciation of evidence. The architecture is a compromise and was understood as one: domestic awards remained exposed to a legality review that international awards seated in India escaped, while the open-ended public policy of the Saw Pipes era was closed down for both. The retention of patent illegality for domestic awards drew a principled objection that has never been answered on its own terms. If review for illegality apparent on the face of the award is an indefensible intrusion into the parties’ bargain, its confinement to domestic awards protects foreign parties from a defect of Indian law while leaving Indian parties exposed to it, a distribution that inverts the usual logic of protecting the domestic litigant. The Law Commission had recommended the ground’s retention for domestic awards on the ground of the ad hoc sector’s uneven quality, in substance a judgment that review must substitute for the quality control that institutions supply elsewhere. The objection and the answer together frame the 2019 amendment’s premise: build the institutions, and the case for the domestic exception falls away. That the institutions were not built is therefore not an isolated failure of the trail; it is the reason the compromise of 2015, designed as a transitional settlement, remains the standing law of challenge. Sections 34(5) and 34(6) added procedural discipline, requiring prior notice to the other party before a challenge is filed and directing endeavours to dispose of the challenge within one year.

Enforcement was where the amendment changed the most lives. Section 36 was rewritten so that the filing of a Section 34 application does not by itself render an award unenforceable; a party seeking a stay must apply separately, and a court granting a stay of a money award does so on conditions akin to those governing a money decree, including deposit. The change sounds procedural and is in fact a transfer of wealth in time: the award debtor lost the free option of freezing enforcement for the life of the challenge, and the award holder gained an asset that could be executed, negotiated against or borrowed upon while the challenge ran. The history of that single change, removed for a class of cases by Section 87 in 2019, defended by the Supreme Court the same year, and qualified for fraud cases in 2021, is the spine of the later trail, and it is told in sequence in the sections that follow.

The procedural discipline of Sections 34(5) and 34(6) deserves separate notice, because it is the part of the design whose operation is least visible in the reports. The prior notice requirement in Section 34(5) was intended to end the practice of challenges filed and kept from the award holder while limitation manoeuvres ran, and the one-year endeavour in Section 34(6) set an expectation for the challenge tier that matched the timeline imposed on the arbitral tier. Courts treated the one-year direction as directory in effect, since the consequence of breach is unstated, but its presence in the text armed case management: challenge petitions began to be listed with a discipline that the pre-amendment docket had not known, and the requirement that the challenge be decided, in the amended Section 34 after 2019, on the record of the tribunal reinforced the direction by shrinking the material a challenge court must master. On the enforcement side, the courts administering the amended Section 36 developed a deposit practice under which a stay of a money award is ordinarily conditioned on deposit of the whole or the greater part of the awarded sum, with departures reasoned from the texture of the challenge. Award holders in infrastructure and public works disputes, for whom the award is frequently the difference between solvency and default, treated the deposit as the amendment’s operative gift, since a deposit converts a frozen claim into secured value even before the challenge is decided. That practice is the background against which the unconditional fraud stay of 2021 must be read: an unconditional stay is exceptional in the post-2015 dispensation precisely because the ordinary stay is bought with money down.

Two extensions of reach completed the 2015 design. A proviso to Section 2(2) extended Sections 9, 27 and clause (a) of Section 37(1) together with Section 37(3) to international commercial arbitration even where the place of arbitration is outside India, unless the parties agree otherwise, reversing for those provisions the territorial exclusion that the Supreme Court had laid down for the pre-amendment Act. And Section 8, governing referral by a judicial authority faced with an action covered by an arbitration agreement, was tightened so that referral follows unless the authority finds prima facie that no valid arbitration agreement exists, with the application to be made not later than the first statement on the substance of the dispute.

One jurisdictional repair completed the court-facing design. The definition of Court in Section 2(1)(e) was substituted so that, in international commercial arbitration, the court for the Act’s purposes is the High Court exercising jurisdiction over the seat, and in other arbitrations the principal civil court of the district, a distribution that matched the new commercial courts architecture of the same legislative season and ended the anomaly of district courts entertaining applications arising from international arbitrations seated in India. The change is minor in the text and consequential in the docket, since the Act’s applications, for interim relief, for appointment support, for challenge and for enforcement, thereafter sorted themselves between the commercial divisions of the High Courts and the district judiciary on a stated principle, and the jurisprudence of the Act accumulated in forums equipped for commercial work.

What did the 2015 amendment extend to arbitrations seated outside India?

The proviso to Section 2(2) carried Sections 9, 27 and parts of Section 37 into international commercial arbitration even where the seat is outside India, unless parties have agreed to the contrary. An Indian party to a foreign-seated arbitration could therefore seek interim protection from an Indian court over assets or evidence in India, reversing the territorial exclusion.

Section 26 and the settlement in BCCI v Kochi Cricket

Every amendment that changes enforcement rights for pending cases writes a dispute about its own reach, and the 2015 amendment wrote its reach in Section 26 of the amendment Act. Section 26 provided that nothing in the amendment Act shall apply to arbitral proceedings commenced before its commencement, unless the parties otherwise agree, while adding that the amendment Act shall apply in relation to arbitral proceedings commenced on or after that date. The text divided the world by the date of the arbitration, and said nothing express about court proceedings, which are the proceedings in which Section 34 and Section 36 actually operate. High Courts divided on whether the improved enforcement regime of the amended Section 36 reached a Section 34 application filed after 23 October 2015 in an arbitration begun before that date.

The Supreme Court settled the question in Board of Control for Cricket in India v Kochi Cricket Partners LLP (2018) 6 SCC 287, decided on 15 March 2018 by a bench of R. F. Nariman and Navin Sinha JJ. The Court held that Section 26, read as a whole, addressed arbitral proceedings, and that court proceedings were governed by the ordinary principle that a procedural amendment applies to proceedings instituted after its commencement. The amended Section 36 therefore applied to Section 34 applications filed on or after 23 October 2015 even where the arbitration itself had begun earlier, with the consequence that the automatic stay died for every challenge instituted after the amendment, whenever the underlying arbitration had started. The judgment is the fixed point of the trail: everything Parliament did in 2019 about applicability was an attempt to displace it, and everything the Supreme Court did in reply was a defence of it.

The stakes in the case were not doctrinal abstractions. The class of pending challenges to which the question applied was dominated by public works and infrastructure awards, in which the award debtor was a public authority and the award holder a contractor whose working capital was locked in the award. In that class, the automatic stay had operated as a financing device for the debtor and a slow strangulation for the creditor, and the difference between the two constructions of Section 26 was measured in the survival of firms. Hindustan Construction itself, the petitioner in the later case, was an infrastructure contractor carrying awards against the Union of India, and the litigation posture of the two cases, contractors invoking the amended regime against public debtors who invoked the older one, explains the alignment of institutional support on each side: industry bodies for the reformed Section 36, and the finance departments of governments for its confinement. The reasoning repays attention because it frames the later constitutional strike. The Court treated the removal of the automatic stay as a matter of the award holder’s right to execute, a right that accrues on the making of the award and is not a mere matter of procedure in the hands of the award debtor. It also recorded, with unusual directness, that a legislative proposal to reverse the position by inserting a Section 87 had been placed before it in the course of argument as a Bill, and that the enactment of such a provision would be contrary to the object of the 2015 reform. The warning was delivered before the provision existed. When the provision arrived the following year, it arrived against a judgment that had already explained, in terms, why its content was objectionable, and the collision that followed was therefore not an accident of drafting but the working out of a disagreement between Parliament and the Court about which of them should fix the temporal reach of the 2015 settlement.

The amendment of 2019: the institutional turn

The 2019 amendment answered a different diagnosis from the one the Law Commission had offered. The High Level Committee to Review the Institutionalisation of Arbitration Mechanism in India, chaired by Justice B. N. Srikrishna and reporting on 30 July 2017, located India’s arbitration problem in its structure rather than in court behaviour: arbitrations were ad hoc, arbitrators were appointed by courts or by one party, institutions were few and weak, and the state, the largest litigant in commercial disputes, had no framework for conducting arbitration as a repeat participant. The Committee recommended a specialist statutory body to promote institutional arbitration, grading of institutions, accreditation of arbitrators, model guidelines for government contracts, and a set of amendments to the 1996 Act. The amendment of 2019, enacted as Act 33 of 2019 with assent on 9 August 2019, adopted the institutional programme in form. Its subsequent history is the record of how little of that programme was switched on.

Part IA, inserted by Section 10 of the amendment Act, created the Arbitration Council of India in Sections 43A to 43M. The Council was to consist of a chairperson who has been a judge of the Supreme Court, or a Chief Justice or judge of a High Court, or an eminent person with expert knowledge in arbitration, appointed by the Central Government in consultation with the Chief Justice of India, together with members drawn from government, the legal profession, arbitral institutions and commerce. Its functions under Section 43D included grading arbitral institutions, accrediting arbitrators through those institutions, framing policy and guidelines for the establishment, operation and maintenance of arbitral institutions, holding training and examinations, maintaining a repository of arbitral awards, and reviewing the grading of institutions. Section 43J tied accreditation to qualifications, experience and norms specified in the Eighth Schedule. The design placed the state at the centre of a market for dispute resolution services, and it drew criticism on exactly that ground, including from commentators who supported institutional arbitration in principle, that a body appointed by the government of the day would grade the institutions before which that government is the most frequent party. That criticism is part of the record of the design; the design itself was never tested, for reasons the next section sets out.

The composition provisions repay a closer look, because they explain both the ambition and the resistance. Section 43C built the Council around a chairperson drawn from the higher judiciary or from eminence in the arbitration field, a member expert in institutional arbitration, ex officio representation from the departments of the Central Government concerned with law and with public expenditure, a part-time member from commerce and industry, and a chief executive drawn from the senior civil service. The grading function in Section 43D was to operate on criteria the Council itself would frame, covering infrastructure, arbitrator quality, case disposal performance and compliance with statutory timelines, and the Council was to maintain an electronic repository of arbitral awards made in India and to review the grading of institutions periodically. Accreditation of arbitrators was to run through the graded institutions against the qualifications in Section 43J and the Eighth Schedule, so that an individual’s route to tribunal appointment in institutional arbitration would pass, at two removes, through a body appointed by the executive. Defenders of the design, including voices in the arbitration institutions that expected to be graded, argued that a market with no quality signal leaves parties at the mercy of reputation networks, and that every mature seat possesses some combination of chartered institutions and professional standards to do the work the Council was designed for. Critics answered that the design confused a standards function with a licensing function, and that grading by a state body in disputes where the state is the dominant party inverts the neutrality the 2015 amendment had legislated for. The disagreement was never resolved on the merits, because the commencement decision was never taken, and the design survives as the fullest statement of a theory of arbitration governance that India enacted and declined to operate.

The appointment scheme was rebuilt to run through the Council’s graded institutions. The substituted Section 11 provided that the Supreme Court and the High Courts would designate arbitral institutions graded by the Council, and that appointments on a party’s default would be made by the designated institution rather than by the court, with the court’s own appointment power surviving only where no graded institution was available. Section 11(6A), the 2015 confinement of the appointing court to the existence of the agreement, was to fall away because the court itself was to fall away from the function. The Eighth Schedule prescribed the qualifications and experience for accreditation as an arbitrator, and its categories, framed around Indian professional and service backgrounds, had the effect of excluding foreign qualified practitioners from accreditation, in a statute whose stated ambition was to attract international arbitration to Indian seats. That schedule became the most criticised text of the amendment, and its removal two years later was the least contested act of the 2021 amendment.

The architecture of the Eighth Schedule explains the force of the criticism. Its categories were drawn from the organised professions and the public services of India: advocates within the meaning of the Advocates Act 1961 with a decade of practice, chartered accountants, cost accountants and company secretaries of comparable standing, officers of the Indian Legal Service and of government legal services, serving and retired officers with legal or technical qualifications in public employment, and a residual category for persons with knowledge and experience in the conduct of arbitration. Each category carried experience thresholds and each was recognisably Indian in its reference points, so that a foreign advocate of thirty years’ standing at a leading international firm, a foreign engineer of international repute, or a retired judge of a foreign commercial court fell outside the schedule’s frame altogether. The schedule’s defenders, a minority in the debate, argued that accreditation regulates a public-facing function and that a seat building its institutions is entitled to anchor eligibility in its own professional systems. The dominant view, expressed by Indian and foreign institutions alike and by the Indian practitioners who compete for international appointments, was that eligibility anchored in nationality of qualification is a barrier dressed as a standard, and that the schedule would have operated to keep international commercial arbitration administered from Singapore and London even where the seat was Indian. The 2024 consultation draft, which proposed that accreditation norms be framed without the schedule’s categories, recorded the official acceptance of that verdict.

The parts of the 2019 amendment that did commence were the working parts. Section 23(4) required the statement of claim and defence to be completed within six months of the arbitrators receiving notice of their appointment. Section 29A was recast so that the twelve-month period for the award runs from the completion of pleadings under Section 23(4) rather than from the tribunal entering upon the reference, and the timeline was withdrawn from international commercial arbitration, for which the obligation became an endeavour to dispose expeditiously; the fee-reduction and substitution machinery of the section was adjusted in the same operation. Section 34 was amended so that a challenge is established on the basis of the record of the arbitral tribunal, replacing the earlier language of furnishing proof and confining the material before the court. Sections 17, 37, 45 and 50 received consequential amendment. Sections 42A and 42B were new. Section 42A imposed a duty of confidentiality on the arbitrator, the arbitral institution and the parties, broken only where disclosure of the award is necessary for its implementation and enforcement. Section 42B protected an arbitrator from suit or other legal proceedings for anything done or intended to be done in good faith under the Act. Both provisions commenced on 30 August 2019 and operate.

The operational consequences of the notified cluster deserve record, since they are the 2019 amendment’s actual legacy in practice. The recast timelines changed tribunal case management: with the award clock running from the completion of pleadings, and pleadings themselves capped at six months from the tribunal’s constitution notice, the front of a domestic arbitration acquired fixed stations, and tribunals began issuing procedural timetables keyed to Section 23(4) as a matter of form. The Section 34 record rule changed challenge practice in a quieter way: counsel compiling a setting-aside petition after August 2019 worked from the tribunal’s record as the closed universe of material, and the older habit of supplementing the record with affidavits of evidence never before the tribunal lost its textual footing, so challenges shortened and their grounds clustered on the face of the award and the procedural record. The confidentiality duty of Section 42A altered drafting practice for arbitration clauses and procedural orders, which began to carry express confidentiality terms aligned to the section, while its enforcement exception ensured that award disclosure in execution proceedings continued without argument. Section 42B’s good faith protection, rarely invoked in reported litigation, performed its work invisibly, in the reduced willingness of disappointed parties to name arbitrators personally in collateral proceedings. None of these effects required the Council, and their accumulation is the answer to a puzzle the trail poses: the 2019 amendment reads in the commentaries as a failure, yet its notified provisions are among the most routinely applied texts in Indian arbitration practice.

Section 42A carries one limit that its early operation exposed. The duty binds the arbitrator, the institution and the parties, and its exception covers disclosure of the award for implementation and enforcement, but the section states no consequence for breach and no mechanism for the publication practices by which institutions elsewhere build a body of guidance: the anonymised release of awards, the reporting of procedural decisions, the statistical disclosure of caseloads. Institutions administering India-seated arbitrations therefore developed their confidentiality terms by contract, layering procedural orders over the statutory floor, and the question whether the statutory duty, read strictly, obstructs the anonymised publication that a maturing seat needs for precedent and for scrutiny was debated in the consultation on the 2024 draft without a settled answer. The point is minor beside the commencement failure, and it is recorded here because it shows the notified 2019 provisions generating the ordinary second-generation questions of a working text, a different order of problem from the first-generation question, whether the text operates at all, that Part IA never escaped.

The amendment’s final operation was its most consequential and its shortest lived. Section 15 of the amendment Act deleted Section 26 of the 2015 amendment Act, and the main Act received a new Section 87, which provided that the 2015 amendments shall not apply to arbitral proceedings commenced before 23 October 2015, nor to court proceedings arising out of or in relation to such arbitral proceedings whenever those court proceedings were commenced, and shall apply only to arbitral proceedings commenced on or after that date and to court proceedings arising from them. The provision was deemed to operate from 23 October 2015. Its effect was to reverse Board of Control for Cricket in India v Kochi Cricket Partners by legislation: the automatic stay, abolished by the amended Section 36 for challenges filed after the 2015 commencement, revived for the entire class of pre-commencement arbitrations, however late their challenges were filed. The Statement of Objects and Reasons defended the change as a clarification of the position the Srikrishna Committee had recommended, on the footing that the 2015 amendment was prospective. The judgment it displaced had considered that footing and rejected it.

The commencement failure: a Part that never began

The 2019 amendment was not brought into force as a whole. By notification S.O. 3154(E) dated 30 August 2019, the Central Government appointed 30 August 2019 as the date of commencement for Sections 1, 4 to 9, 11 to 13 and 15 of the amendment Act. Those sections produced the amendments to Sections 17, 23, 29A, 34, 37, 45 and 50, the insertion of Sections 42A and 42B, and the insertion of Section 87 with the deletion of Section 26 of the 2015 Act. No notification followed for Sections 2, 3, 10 and 14 of the amendment Act. Section 2 contained the definition amendments, including the definition of an arbitral institution by reference to designation. Section 3 contained the substituted Section 11, the institutional appointment scheme. Section 10 contained Part IA and with it the Arbitration Council of India. Section 14 contained the Eighth Schedule. The commencement record, and not the enacted text, is therefore the accurate map of the 2019 amendment: the procedural amendments live, the institutional amendments do not.

The consequences are concrete and checkable. No chairperson or member of the Arbitration Council of India has been appointed under Part IA, because the Part that would receive the appointment is not in force. No arbitral institution has been graded by the Council, and no arbitrator has been accredited under Section 43J. The Supreme Court and the High Courts have made no designations under the substituted Section 11, because that section is not in force either; appointments on default continue to be made by the courts under Section 11 in the form given to it in 2015, confined by sub-section (6A) to the existence of the arbitration agreement. Secondary literature that describes the Council as a functioning regulator of Indian arbitration, of which there is a quantity, describes a body that exists only in the uncommenced portion of an amendment Act. The position recorded in the consultation exercise of 2024, which proposed redesigning the Council’s composition and functions before it had ever sat, confirms the record: the draft of a fourth amendment proceeded on the express footing that the institutional settlement of 2019 remained unbuilt.

The pattern is not unique to arbitration, and this series has recorded it in fields far from commercial law, where commencement was staggered for years and in part never completed, as the account of the labour codes and their implementation delay sets out for the consolidation of the labour statutes. What distinguishes the arbitration instance is the position of the uncommenced matter in the scheme. A deferred schedule defers a detail. The deferred Part IA deferred the institution on which the amendment’s other designs depended: the appointment scheme required graded institutions, the grading required the Council, and accreditation required both. Commencement of Sections 2, 3, 10 and 14 was thus a single decision with four labels, and it was never taken. Two explanations are offered in the literature and both deserve statement. The explanation favourable to the government is that grading and accreditation machinery requires groundwork, consultation with institutions and courts, and staffing, and that commencing the Part before that groundwork existed would have produced a hollow regulator and litigation over its first acts. The critical explanation is that a reform whose centre is a state-appointed grading body for a market in which the state is the largest disputant met resistance from the institutions, the bar and the judiciary whose functions it displaced, and that indefinite non-commencement was the politically cheapest resolution of that resistance. The record supports one proposition beyond argument: the gap between enactment and commencement, across the whole period since August 2019, is measured in years and not in the months that groundwork would explain, and the 2024 consultation proposed to redesign the body rather than to commence it, which is the treatment reserved for designs a government no longer intends to build in their enacted form.

Institutional arbitration in India did not wait for Part IA. The decade of the trail saw institutions built by instruments other than the principal Act. The New Delhi International Arbitration Centre was established by its own statute in 2019 to take over the assets and functions of the earlier international centre in the capital, and was renamed the India International Arbitration Centre by an amendment of 2022, a free-standing institutional vehicle that owes nothing to the Arbitration Council design. Private and court-annexed institutions expanded in the commercial centres, the Mumbai Centre for International Arbitration building an administered caseload under its own rules from 2016, and court-annexed centres in Delhi, Madras, Bombay and elsewhere regularising administered arbitration under the supervision of the High Courts. Empirical accounts of Indian arbitration in the period record growth concentrated in these institutions and in ad hoc arbitration under the reformed Act, with the administered share rising from a low base without any grading or accreditation by a statutory council. The comparison is the sharpest commentary on the 2019 design’s fate: the function the Council was created to perform, the accreditation of a trustworthy institutional sector, was pursued by the market, by the courts and by a separate statute, while the Council’s own Part remained uncommenced. A grading regime imposed on a sector that had organised itself without one would have arrived, if commenced late, as an answer to a question the sector had stopped asking.

Hindustan Construction: the Court removes Section 87

Section 87 operated for under three months. On 27 November 2019, a bench of R. F. Nariman, Surya Kant and V. Ramasubramanian JJ decided Hindustan Construction Co Ltd v Union of India, reported as 2019 SCC OnLine SC 1520 and in the Supreme Court Cases as (2020) 17 SCC 324, and struck Section 87 down. The deletion of Section 26 of the 2015 amendment Act, which was Section 87’s companion and had no other work to do, fell with it. The consequence was that the 2015 amendments, including the amended Section 36, applied to court proceedings initiated on or after 23 October 2015 irrespective of when the underlying arbitral proceedings had commenced, which is to say that the position laid down in Board of Control for Cricket in India v Kochi Cricket Partners governed again, and governs in the absence of any later legislative reversal of the point.

The ground was Article 14 of the Constitution, applied through the doctrine of manifest arbitrariness. The Court reasoned in steps. The 2015 amendment had removed the automatic stay because the stay had become an instrument of delay that harmed award holders and the public fisc without protecting any legitimate interest of award debtors, who retained a full right to apply for a stay on conditions. Section 87 restored the stay for a large and closed class of cases defined by a date, and did so retrospectively, reviving stays in proceedings where challenges had been filed in reliance on the amended law and where courts had already refused or conditioned stays under it. No rationale connected the classification to the object of the parent reforms; the Statement of Objects and Reasons invoked the Srikrishna Committee, but the Committee had reported in July 2017, before the Kochi Cricket judgment, and had not addressed the construction the Supreme Court had placed on Section 26 the following year. A provision that resurrects a mischief Parliament had removed, for no reason the record supplies, and that unsettles accrued rights in pending proceedings, was held to be capricious and without an adequate determining principle, and therefore manifestly arbitrary.

The distinction the judgment preserves is the one Indian validation law has always drawn, and it explains why the strike did not unsettle legislative power over the field. A legislature dissatisfied with a judicial construction may legislate against it, and routinely does, by altering the legal basis on which the construction rests: changing a definition, supplying a deeming provision with a rationale, restructuring the right itself. Such legislation is reviewed for competence and for constitutional limits, and within those limits it displaces the judgment for the future and, where the text so provides, for the past. Section 87 did none of that work. It left the amended Sections 34 and 36 untouched, left the Kochi Cricket reasoning about accrued enforcement rights unaddressed, and simply directed that the earlier law be treated as governing a class of cases, on a justification that referred to a committee report predating the judgment it displaced. The Court’s holding, confined to that configuration, leaves Parliament’s ordinary instruments intact, and the 2021 amendment’s design shows the lesson absorbed: the fraud proviso does not direct courts to disregard a construction, it creates a new substantive ground of relief, with its own threshold and its own rationale stated in the parliamentary record, and applies that ground retrospectively by express words. Retrospection alone was never the vice. The vice was retrospection as a bare instruction to treat a condemned position as law.

The case is the trail’s hinge because it changed the character of every later move. A legislature confronted with a construction it dislikes may ordinarily legislate the construction away, and courts accept the displacement when the legislature removes the basis of the judgment rather than re-enacting the condemned position. Section 87 attempted the second course without the first: it did not alter the reasoning about accrued enforcement rights, and it restored the very consequence the reasoning had condemned. After the strike, the field was left in a specific configuration. The automatic stay was gone for post-commencement court proceedings and was not coming back by simple re-enactment. Any further protection for award debtors had to be framed as a judicially administered, case-specific relief, which is the form the protection for fraud cases took in 2021. The judgment also stands as the only instance in the arbitration statute’s history of an amendment provision being invalidated on a substantive equality ground, a point that frames the verdict offered at the end of this article, and it belongs to the wider pattern, traced in the field guide to the judicial system laws of India, of the constitutional courts supervising the terms on which commercial justice is organised.

The doctrine the Court applied deserves its own sentence in the trail. Manifest arbitrariness entered the law of legislation review through the plurality opinions in Shayara Bano v Union of India (2017) 9 SCC 1, which recognised that a statute may be struck down under Article 14 where it is capricious, irrational or without an adequate determining principle. Its application to a fiscal-adjacent commercial provision in Hindustan Construction was the doctrine’s first major use against an economic statute, and it was carefully bounded: the Court did not hold that Parliament lacks the power to fix the applicability of an amendment, a power legislatures exercise routinely, but that a provision which resurrects a removed mischief, unsettles accrued rights, and offers no rationale that survives contact with the record is not an exercise of that power in a constitutionally intelligible form. Award debtors who had obtained the benefit of the revived stay in the provision’s short life lost it retrospectively, and enforcement proceedings across the country resumed on the footing of Kochi Cricket, with deposits ordered in cases where the stay had operated unconditionally. The episode left a mark on drafting practice beyond arbitration: applicability clauses in later commercial amendments are drawn with an evident care to state their rationale and their treatment of accrued rights on the face of the objects, a discipline the arbitration trail taught at some cost.

The amendment of 2021: a fraud exception and a schedule removed

The third amendment arrived as the Arbitration and Conciliation (Amendment) Ordinance 2020, promulgated on 4 November 2020, and was enacted as the Arbitration and Conciliation (Amendment) Act 2021 (Act 3 of 2021), passed by the Lok Sabha on 12 February 2021 and by the Rajya Sabha on 10 March 2021, with assent following in March 2021 and the commencement of the Act deemed to be 4 November 2020, the date of the ordinance. Its political driver was a stated concern about awards procured by fraud or corruption, including awards in favour of parties who had obtained them through corrupt means and who then pressed for enforcement and for the withdrawal of attached sums while the challenge was pending. The enacted remedy was narrow in its trigger and absolute in its effect.

The operative change is a further proviso to Section 36(3), the provision under which a court granting a stay of a money award imposes conditions. The proviso directs that where the court is satisfied that a prima facie case is made out that the arbitration agreement or the contract which is the basis of the award, or the making of the award, was induced or effected by fraud or corruption, the court shall stay the award unconditionally until the challenge under Section 34 is disposed of. Three features define the provision. The threshold is a prima facie case, a provisional assessment on the material before the court at the stay stage, and not a finding. The consequence is mandatory and unconditional: on satisfaction of the threshold the court has no discretion to exact a deposit or security, which distinguishes the fraud stay from every other stay under the section. And the reach is retrospective in the strongest form the statute book knows: the proviso is deemed to have been inserted with effect from 23 October 2015, and an explanation declares that it applies to all court cases arising out of or in relation to arbitral proceedings, irrespective of whether the arbitral or court proceedings were commenced before or after the commencement of the amendment. The drafting was noticed in Parliament, where members asked why a remedy for a demonstrated concern required a deemed insertion reaching back across five years of enforcement orders; the government defended the reach on the ground that a fraud-tainted award should never have been enforceable at any point after the 2015 settlement.

The strongest case for the provision is the case its movers made. An award that is the fruit of fraud or corruption is not an ordinary award with an ordinary defect; enforcement of such an award while a challenge runs can dissipate the money beyond recovery, and the ordinary stay machinery, conditioned on deposits that an award holder under a tainted award can meet precisely because the award is large, is an inadequate protection. Fraud unravels advantages obtained by it, the prima facie threshold confines the indulgence to cases where a court sees substance at first sight, and the stay lasts only until the Section 34 challenge is disposed of, so the award holder under a genuine award loses time rather than the award. The strongest case against the provision is the case made in the parliamentary debate and in commentary by arbitration practitioners. The threshold is low relative to the consequence: an unconditional stay with no deposit, obtained on a prima facie showing, reproduces for the pleading party the substance of the automatic stay abolished in 2015, and the class of cases is defined by an allegation that losing parties have every incentive to make. Retrospection across the whole period since 2015 reopens enforcement positions settled under the law as the Supreme Court had declared it, and the fraud exception, designed for the corrupt award, is available in every large commercial dispute in which fraud is pleaded as a matter of routine advocacy. A verdict between these cases is possible on the provision’s own terms. The trigger is defensible where the fraud alleged goes to the making of the award or to the contract that is its basis, which the text requires, and the mischief predicted against it depends on courts applying the prima facie threshold loosely, which is a question of administration rather than of text. The retrospection is harder to defend: no case was offered in the debate for disturbing stays refused and executions completed between 2015 and 2020 in matters with no fraud finding against them, and the explanation’s sweep catches those matters by its terms.

The parliamentary passage of the 2021 amendment recorded both cases in the legislators’ own terms. Members opposing the Bill objected to the ordinance route for a measure that reached back five years, to the breadth of a fraud plea available in every commercial dispute, and to the absence of any deposit condition, which they characterised as a stay for the asking in the hands of a party willing to plead fraud with colourable material. Members supporting the Bill answered from a record of enforcement experience in public procurement and infrastructure, where awards of great size, later attacked as procured, had been executed against the exchequer while challenges ran, and where the practical irreversibility of a completed execution made the ordinary stay machinery an inadequate protection for public money. The government’s defence of the deemed insertion date was that the 2015 settlement itself had been premised on honest awards, and that a proviso protecting the settlement against its own abuse belonged, in logic, to the settlement’s own date. Critics of that argument, in the debate and in the commentary that followed, answered that logic of that kind proves too much, since every protective amendment could claim the date of the settlement it qualifies. The amendment passed with its retrospectivity intact, and its administration was left, as the opponents predicted and the supporters accepted, to the discipline of the courts applying the prima facie threshold case by case.

The amendment’s second operation closed the Eighth Schedule controversy. Section 43J was substituted so that the qualifications, experience and norms for accreditation of arbitrators shall be such as may be specified by the regulations, and the Eighth Schedule to the principal Act was omitted. The criticism that produced the change had come from two directions and is worth stating in the voices that made it. The international arbitration community, speaking through institutions and practitioners whose participation Indian seats were competing to attract, objected that a schedule of qualifications framed around Indian legal and administrative service excluded foreign advocates, engineers and accountants from accreditation and signalled a closed market. Domestic critics objected on the connected ground that accreditation by schedule would fossilise eligibility while arbitration’s subject matter diversified. The government, in making the change, accepted the substance of both objections, and chose omission over redrafting: the schedule went, the accreditation power remained in Section 43J in a regulation-based form, and since the Part containing Section 43J has never been commenced, the omission and the substitution alike operate on a framework that is not in force. The schedule was therefore removed from a text that no arbitrator has ever been accredited under, a curiosity that the table in this article records in its own row.

Why does Section 43J appear in the Act when the Council does not exist?

Section 43J sits in Part IA, and Part IA has never been brought into force. The section appears in the printed Act because the 2019 amendment enacted it and the Gazette records enactment, while commencement, the step that makes a provision operative, was never notified for the Part. Section 43J is enacted law awaiting a commencement that has not come.

It remains to record what the 2021 amendment did not do, because the point decides what law a reader of summaries might otherwise misstate. It did not omit or repeal Section 87. Section 87 had been struck down by the Supreme Court in November 2019, so there was no operative provision for the 2021 Act to remove, and the Act neither revived it nor re-enacted its content; the Kochi Cricket settlement, as restored by Hindustan Construction, was left standing. The contrast with the treatment of the Tribunals Reforms legislation, examined in the article on the Tribunals Reforms Act 2021, is instructive: there, provisions of an ordinance that the Supreme Court had ruled against were re-enacted in altered institutional settings, and the constitutional contest continued. In arbitration, Parliament accepted the strike and legislated on the adjacent ground of fraud-specific relief. Whether that restraint was constitutional prudence or exhaustion with the subject, its effect is that the applicability law of the arbitration statute is judge-made to a degree unusual in a codified commercial field.

The drafting of the proviso’s reach repays a final look, because it is the most technically accomplished piece of temporal drafting in the trail. A simple commencement on the ordinance date of 4 November 2020 would have protected the exchequer only against enforcement steps taken after that date, leaving award holders in the five preceding years, the entire working life of the amended Section 36, free to execute awards of the very class the amendment condemned. A repeal and re-enactment of Section 36 with a fraud exception would have reopened the whole enforcement settlement, including the Kochi Cricket construction, to fresh contest. The device chosen, a proviso deemed inserted from the settlement’s own first day, accompanied by an explanation addressed to the courts’ case load rather than to arbitrations, threaded both difficulties: it located the exception inside the provision it qualifies from the moment that provision began, and it spoke to court cases because enforcement lives in court cases, whatever the date of the arbitration beneath them. The technique has its cost, which the opponents in the debate named: executions completed in reliance on the law as declared are exposed to challenge on a threshold far below a finding of fraud. The text accepts that cost openly rather than concealing it, and the openness is the difference, in the trail’s own constitutional grammar, between this retrospectivity and the retrospectivity that failed in 2019.

The administration of the fraud proviso in its first years supplied the evidence on which the provision will eventually be judged. High Courts approached the prima facie threshold as a real filter rather than a formality: the fraud or corruption alleged must be shown, on material and not on assertion, to touch the arbitration agreement, the underlying contract or the making of the award itself, and allegations that amount to a merits grievance dressed in the language of fraud, an erroneous appreciation of evidence, a disputed interpretation, a disappointed expectation of the outcome, have been held insufficient to trigger the unconditional stay. Courts also policed the boundary between the proviso and the ordinary stay jurisdiction, treating an applicant who fails the threshold as remitted to the conditioned stay under the main provision, with deposit, rather than as entitled to an unconditional indulgence by a lower route. The early pattern therefore cuts against the strongest form of the prediction made against the provision in 2021: the plea is made frequently, because the incentive to make it is built into the design, and it succeeds in a minority of the cases in which it is made, because the threshold, administered as written, requires a prima facie case on documents that courts can test at the stay stage. Whether that discipline holds as the provision’s case law accumulates is an empirical question the trail leaves open, and it is the question on which the 2021 amendment’s reputation will turn.

The narrowed challenge ground in the courts after 2015

The amendments supply the text of the trail; the courts supplied two of its terminal points. The first, on Section 87, is recorded in its own section. The second concerns the operation of the narrowed Section 34 in practice. In Ssangyong Engineering and Construction Co Ltd v National Highways Authority of India (2019) 15 SCC 131, the Supreme Court applied the post-2015 architecture to a highway construction award and confirmed the settlements the amendment had intended: the public policy ground is confined to the categories in the explanation, a contravention of the fundamental policy of Indian law does not entail a review on the merits, and patent illegality under Section 34(2A), available only outside international commercial arbitration, is a ground that goes to the root of the matter and does not permit a court to reappreciate evidence or to substitute its reading of a contract for the tribunal’s where the tribunal’s reading is a possible one. The judgment matters to the trail because it completed the 2015 design from the judicial side: Parliament narrowed the text, and the Court declined to widen it by construction, so the Saw Pipes expansion stands reversed by the combination of the two acts.

The narrowing had a judicial preface. In Associate Builders v Delhi Development Authority (2015) 3 SCC 49, decided months before the amendment commenced, the Supreme Court restated the public policy categories restrictively and warned against merits review under the cover of the ground, a judgment that signalled the Court’s own retreat from the Saw Pipes breadth and that the 2015 explanation then codified. The sequence matters for a fair account of the trail: the restriction of Section 34 was a joint product, with the judiciary retreating first in doctrine and Parliament then fixing the retreat in text, and Ssangyong confirming the text against attempts to reopen it in construction.

Emergency arbitration travelled the opposite route, from practice into case law and from case law towards the statute book, without ever entering the enacted text. Institutional rules available in India-seated arbitration permitted a party to seek urgent interim relief from an emergency arbitrator before the tribunal’s constitution, and the question whether such an order binds in Indian law reached the Supreme Court in Amazon.com NV Investment Holdings LLC v Future Retail Ltd, AIR 2021 SC 3723. The Court held that an emergency arbitrator’s order under the rules of an institution, in an arbitration seated in India, is an order within Section 17(1) and enforceable under Section 17(2), and that no appeal lies against its enforcement under Section 37, reasoning that party autonomy in adopting institutional rules carries the emergency mechanism with it. The position for foreign-seated arbitrations was left on the different footing the territorial provisions supply. The development is part of the amendments story because it exposed, in a high-value retail dispute conducted in public view, the gap between the statute’s interim relief scheme and the machinery commerce had adopted for itself, and the 2024 draft’s proposed Section 9A is the trail’s legislative answer to that exposure.

The judicial terminus of the appointment line came from a Constitution Bench. In Central Organisation for Railway Electrification v M/s ECI SPIC SMO MCML (JV), 2024 INSC 857, decided on 8 November 2024 by five judges, the Court addressed clauses in public sector contracts under which one party, a railway organisation, maintained a panel of arbitrators from which the other party was required to choose, and under which appointments could be made unilaterally. The majority, speaking through Chandrachud CJI with Pardiwala and Misra JJ, held that equal treatment of parties at the stage of appointment is a component of the arbitration agreement’s validity under the Constitution’s equality guarantee as much as under the Act, that a clause permitting one party to appoint a sole arbitrator unilaterally, or to confine the counterparty’s choice to a panel it curates, fails that standard in the public sector context, and that the principle in the amended Section 12(5) and the Seventh Schedule informs the reading. Roy and Narasimha JJ delivered separate opinions. The operative directions were made prospective for appointments made after the judgment, in the terms recorded at paragraph 170.7 of the report, expressed for three-member tribunals. For the trail, the decision closes the arc that Section 12 and the schedules opened in 2015: independence doctrine moved from disclosure, to ineligibility, to a constitutional rule about who may compose the pool from which a tribunal is drawn, and it did so by judgment because the appointment provisions that might have carried the question into an institutional framework, those of 2019, were never commenced.

Enacted, notified, struck down: the trail in one view

The table below records each significant change in the trail against the three fates an Indian statutory change can meet: commencement, continued survival, or removal by omission or by a court. It is the article’s consolidation device, and readers who use the trail for a single fact are directed to the row that carries it, since the same provision name appears at several rows’ remove from its fate. The rows on Part IA and the Eighth Schedule state the position that the commencement sections of this article establish: enactment in 2019, no commencement on any date, and, for the schedule, omission in 2021.

Change	What it provided	Enacted	In force or fate
Section 11(6A) confinement	Court confined to examining the existence of the arbitration agreement on appointment	Amendment of 2015	In force 23 October 2015; survives, the institutional replacement never having commenced
Section 12 schedules	Disclosure duties, Fifth Schedule guidance, Seventh Schedule ineligibility under Section 12(5)	Amendment of 2015	In force 23 October 2015; survives
Section 9(3) bar	Court not to entertain interim applications after the tribunal is constituted, unless the Section 17 remedy is inefficacious	Amendment of 2015	In force 23 October 2015; survives
Section 17(2) status	Tribunal interim orders deemed orders of the court for enforcement	Amendment of 2015	In force 23 October 2015; survives
Section 29A timeline	Award within twelve months of the tribunal entering upon the reference, six months more by consent	Amendment of 2015	In force 23 October 2015; recast in 2019
Section 29B fast track	Decision on documents within six months by agreement of the parties	Amendment of 2015	In force 23 October 2015; survives
Section 31A costs	Costs ordinarily follow the event, conduct weighed	Amendment of 2015	In force 23 October 2015; survives
Section 34 explanation and Section 34(2A)	Public policy confined to fraud or corruption, fundamental policy, morality or justice; patent illegality for domestic awards only	Amendment of 2015	In force 23 October 2015; survives
Section 36 enforcement	Filing a challenge no longer stays enforcement; stay by separate application on conditions	Amendment of 2015	In force 23 October 2015; qualified by the fraud proviso of 2021
Section 2(2) proviso	Sections 9, 27 and part of Section 37 extended to foreign-seated international commercial arbitration	Amendment of 2015	In force 23 October 2015; survives
Part IA, Sections 43A to 43M	Arbitration Council of India, grading of institutions, accreditation of arbitrators	Amendment of 2019	Notified on no date; never in force
Institutional appointment scheme	Substituted Section 11 routing appointments through graded institutions designated by the courts	Amendment of 2019	Notified on no date; never in force
Eighth Schedule	Qualifications and experience for accreditation as arbitrator	Amendment of 2019	Never in force; omitted by the amendment of 2021
Section 23(4) and Section 29A recast	Pleadings within six months; twelve months runs from completion of pleadings; international commercial arbitration on endeavour	Amendment of 2019	In force 30 August 2019
Sections 42A and 42B	Confidentiality duty on arbitrator, institution and parties; good faith protection for arbitrators	Amendment of 2019	In force 30 August 2019
Section 87	2015 amendments confined to arbitrations commenced on or after 23 October 2015, court proceedings whenever begun	Amendment of 2019	In force with retrospective effect 30 August 2019; struck down 27 November 2019
Deletion of Section 26 of the 2015 Act	Removal of the 2015 amendment’s own applicability clause	Amendment of 2019	Struck down with Section 87, 27 November 2019
Section 36(3) fraud proviso	Unconditional stay where a prima facie case of fraud or corruption in the agreement, the contract or the making of the award is made out	Amendment of 2021	Act deemed in force 4 November 2020; proviso deemed inserted with effect from 23 October 2015
Section 43J substituted	Qualifications, experience and norms for accreditation to be specified by regulations	Amendment of 2021	In force with the 2021 Act; operative framework in Part IA remains uncommenced

The draft of 2024: a fourth rewrite in consultation

The trail’s next document is not an Act. In June 2023 the Central Government constituted an expert committee under the chairmanship of Dr T. K. Viswanathan, a former Secretary of the Department of Legal Affairs and former Secretary General of the Lok Sabha, to examine the working of the 1996 Act and recommend reforms. The committee reported in February 2024. On 18 October 2024 the Department of Legal Affairs released the Draft Arbitration and Conciliation (Amendment) Bill 2024 for public consultation, circulating the draft text and the committee’s reasoning as office documents that practitioners read against the amended Act, a comparison best worked through in an office file viewer for Excel, DOCX and PPTX files rather than from summaries of either text. The document is a consultation draft: it has not been enacted, and the positions it proposes are proposals, recorded here because the brief of this article includes the state of the reform agenda and because the draft’s treatment of the earlier amendments is evidence of how the trail is officially read.

The draft proposes to rename the statute the Arbitration Act 1996, deleting conciliation from the title and the Act, on the ground that conciliation received its own statutory home in the Mediation Act 2023. It proposes statutory recognition of emergency arbitration through a Section 9A under which an emergency arbitrator appointed before the constitution of the tribunal may grant interim relief, building on the position the Supreme Court had reached by interpretation in Amazon.com NV Investment Holdings LLC v Future Retail Ltd, AIR 2021 SC 3723, where an emergency arbitrator’s order under institutional rules in an India-seated arbitration was held enforceable as an order under Section 17. It proposes an appellate arbitral tribunal mechanism as an opt-in alternative to a Section 34 challenge for parties who prefer a second arbitral tier over a court, associated timelines for appointment and challenge, a definition of arbitration that includes proceedings conducted wholly or partly by audio-visual electronic means, and a restated Arbitration Council with altered composition and a narrower role, which is the draft’s implicit judgment on the uncommenced design of 2019: the Council returns as a proposal, changed, because the enacted version was never brought to life.

The appellate tribunal proposal is the draft’s boldest element and its most debated. Under the proposed Section 34A, parties who have agreed in writing to the mechanism may take an award to an appellate arbitral tribunal constituted under institutional rules instead of to a court under Section 34, and the appellate tribunal may confirm, vary or set aside the award on grounds mirroring the challenge grounds, with its decision enforceable as an award. Supporters, including the committee whose report underlies the draft, present the device as an answer to the challenge tier’s duration in court and to the loss of party control at the one stage of the process where control matters most. Critics answer that a second arbitral tier doubles the cost of finality, that parties who lose at the appellate tier will seek the court’s door by whatever route remains, reproducing the litigation the device exists to displace, and that opt-in drafting in standard-form contracts will make the election for parties who never weighed it. The consultation closed without a governmental statement of which view prevailed, and the proposal stands as the clearest available indicator of the direction a fourth amendment would take if introduced. Practitioner responses to the consultation divided along a familiar line, between those who regard the appellate tribunal and emergency provisions as overdue regularisations of practice and those who regard each added tier and each added discretion as a further invitation to the delay the statute exists to defeat.

The renaming proposal carries a substantive tail that illustrates the draft’s consolidation logic. Conciliation occupies Part III of the 1996 Act, Sections 61 to 81, a self-contained code for a consensual process that parties may invoke with or without a dispute in being. The Mediation Act 2023 built a parallel and more detailed statutory home for mediated settlement, covering much of the ground Part III occupies, and the draft’s removal of conciliation from the arbitration statute is an acknowledgement that a statute named for adjudicative processes gains clarity by shedding a consensual one whose regulation lives elsewhere. The change is nominal in form and architectural in effect: the statute that began life in 1996 as a consolidation of arbitration, conciliation and enforcement of foreign awards would stand, under the draft, as a single-subject arbitration code, with enforcement of foreign awards retained as its second Part and conciliation retired to the mediation framework. Whether Parliament, when the Bill is introduced, will accept the tidiness is a question the consultation left open, since Part III retains users in trade and community dispute practice who value a conciliation route unconnected with the mediation statute’s machinery.

Which version of the Act governs a given dispute

The trail leaves the working statute in layers, and a reader matching a dispute to its governing text should work through the layers in order. The account that follows is legal information about which text applies, not advice on any live dispute. First, the principal Act is the 1996 Act in the form given by the notified amendments: the 2015 amendment in full, the notified parts of the 2019 amendment, and the 2021 amendment in full. The unnotified parts of 2019, Part IA, the institutional appointment scheme and the definitions that serve them, form no part of the working text for any dispute of any date. Second, appointments are made under Section 11 as the 2015 amendment left it, by the Supreme Court in international commercial arbitration and by the High Courts otherwise, confined to the existence of the arbitration agreement, and subject after November 2024 to the Constitution Bench ruling on unilateral and panel-based appointment clauses in the public sector. Third, enforcement of an award under challenge follows the amended Section 36: no automatic stay, a separate application for stay, conditions including deposit for money awards, and the fraud exception of 2021 under which an unconditional stay must follow a prima facie showing that the agreement, the underlying contract or the making of the award was induced or effected by fraud or corruption, a provision whose reach extends to court cases whenever the arbitration began. Fourth, the applicability of the 2015 settlement to court proceedings is governed by the Kochi Cricket construction as restored in Hindustan Construction: court proceedings instituted on or after 23 October 2015 attract the amended regime. Fifth, timelines run under Section 29A as recast in 2019, twelve months from the completion of pleadings in domestic arbitration, with the consent and court extension machinery, and with international commercial arbitration governed by the endeavour standard.

A third position, recurring in practice, tests the reach provisions rather than the amendments’ dates. A party to an arbitration seated in Singapore, holding assets or facing dissipation of assets in India, invokes the proviso to Section 2(2) as the 2015 amendment left it: Section 9 interim protection is available from an Indian court notwithstanding the foreign seat, unless the parties have agreed to exclude the Part I provisions that the proviso carries across, and standard institutional clauses are examined for that exclusion with a care the pre-2015 law did not require. The same party cannot obtain from an Indian court an order enforcing an emergency arbitrator’s direction made abroad, the Amazon.com NV Investment Holdings position having been laid down for India-seated proceedings, and the draft Bill of 2024, in proposing Section 9A, addressed the domestic side of that line without altering the territorial one.

Two worked positions show the layers in operation. A contract arbitration begun in 2014, in which the award was made in 2016 and challenged in the same year, sits in the class over which the applicability war was fought: the arbitration predates the 2015 commencement, the court proceedings postdate it. Under the Kochi Cricket construction as restored by Hindustan Construction, the amended Section 36 governs the enforcement, so no automatic stay arose from the filing of the challenge, and any stay was a matter of application and conditions; had the same challenge been pending during the short life of Section 87 between August and November 2019, the automatic stay applied in that window and fell with the strike. If fraud in the making of the award is pleaded in that challenge, the 2021 proviso is available, its deemed insertion reaching the case notwithstanding its age. An arbitration begun in 2022 under a public sector panel clause stands at the other end of the trail: appointment runs under the 2015 form of Section 11, the panel arrangement is exposed to the Constitution Bench ruling of 2024 on equality at the appointment stage, the award must meet the recast Section 29A clock, confidentiality under Section 42A attaches, and enforcement proceeds under the amended Section 36 with its fraud exception, all of it administered by courts and institutions rather than by the Council that the 2019 text promised and the commencement record withheld.

What the trail shows about ordinances in arbitration reform

Two of the three amendments began as ordinances, in October 2015 and November 2020, and the 2015 ordinance’s own date became the deemed commencement date around which the applicability litigation of the following six years turned. The ordinance route settled the calendar of this field more decisively than any debate in either House.

The executive’s parallel retreat: arbitration clauses in government contracts

No account of the trail is complete without the development that ran against it. The state is the largest party to arbitration in India, and the executive’s own contracting policy moved, across the same decade, in a direction the amendments do not predict. Guidelines issued by the Department of Expenditure in June 2024 on the use of arbitration in domestic public procurement advised government entities and their undertakings to restrict arbitration clauses to disputes below a stated value threshold, recorded at ten crore rupees, and to take reasoned decisions, approved at senior level, before agreeing to arbitration above it, preferring mediation and the courts for large public contracts. The stated grounds were the government’s experience of the process: the cost of arbitration in high-value disputes, the quality and consistency of outcomes, the frequency with which awards against the state were challenged by the state itself, and the time consumed across the arbitral and challenge tiers together. The guidelines do not amend the statute and bind no private party, and the arbitration community criticised them as a retreat from the policy the amendments embody, pointing out that the state’s challenge rate against awards it loses is a choice of the state and not a defect of the process. Their significance for the trail is diagnostic. The amendments of 2015 to 2021 were drafted on the premise that better statutory machinery would deepen the state’s commitment to arbitration as its default commercial forum. The 2024 guidelines record the principal user of that machinery concluding, from its own ledger, that the machinery serves it badly at scale. A fourth amendment drafted in that climate is likelier to discipline the state’s own participation, through the institutional and appellate devices the draft proposes, than to repeat the promotional ambitions of 2019, and the trail’s final document should be read with that change of premise in view.

The guidelines also expose a structural feature of the field that the amendments never addressed. Arbitration reform in India has been legislated as if the system’s users were symmetric private parties choosing a forum, while its dominant user is a monopsonist disputant whose procurement contracts write the clauses, whose officers decide whether to challenge awards, and whose challenge behaviour manufactures much of the court load the reforms exist to reduce. The 2015 amendment’s deposit discipline worked on that user as much as on any other, and its success in enforcement is in part the success of a rule the state could not exempt itself from. The institutional amendment of 2019, by contrast, asked that same user to build and submit to a grading authority over its own disputes, an act of self-discipline of a different order, and the commencement record shows the order was not given. The executive guidelines of 2024, retreating from arbitration at the top of the value range while the statute promotes it, complete the picture of a state arguing with itself in three instruments at once, and they are the context in which the next amendment, whenever introduced, will be received by the courts and the market it governs.

The verdict on the trail

The question the stance of this series requires to be answered is whether the amendments made India a more reliable seat for arbitration, and the trail permits a differentiated verdict. The amendment of 2015 did the greater part of what it promised. The appointment gate narrowed, enforcement detached itself from challenge, public policy closed, costs began to follow the event, and the courts, in Ssangyong and in the appointment cases, enforced the narrowing rather than evading it. Its failures were of calibration rather than of direction, and its timeline mechanism required the recasting it received in 2019. The amendment of 2019 is the trail’s cautionary document: a reform cannot be evaluated by its enacted text when its operative centre was never commenced, and an institutional revolution that depends on a commencement notification is a press release until the Gazette carries the notification. The amendment of 2021 removed, rightly and without serious opposition, a qualification schedule that contradicted the statute’s own ambitions, and enacted, in its fraud proviso, a protection whose justification in the corrupt case is real and whose routine pleading in ordinary cases is the risk the administering courts hold. The deeper pattern is the one this article opened with. India amended its arbitration statute three times in six years and left the most ambitious of the three amendments unbuilt, so the working law of arbitration in India is the 2015 settlement as the courts have construed it, maintained against a legislative reversal by a constitutional judgment, and awaiting a fourth amendment whose draft proposes to begin the institutional project again under a shorter name.

The comparison with the seats against which Indian arbitration measures itself is instructive, provided it is kept at the level of method. The leading seats built their positions over decades by accumulating institutional capital: rules revised in cycles, courts issuing practice directions within stable texts, and a professional community whose standards bodies grew out of practice. Textual amendment in those seats is infrequent and, when it comes, conservative, because the system’s credibility rests on the predictability of the framework within which parties contracted. India’s trail runs the other way: three amendments and one draft in a decade, an enacted council that never sat, a qualification schedule enacted, condemned and omitted without operating for a day, and an applicability regime for the central enforcement question settled by a constitutional strike rather than by legislative drafting. None of this made arbitration in India unusable; the caseload grew throughout the period, and the 2015 settlement gave practitioners a workable statute. The verdict the trail supports is narrower and more useful than a ranking of seats. Reform by repeated amendment is a weak instrument for building trust in a dispute system, because each reopening of the text reprices every arbitration clause written under the last version. The amendments that endured are the ones that changed incentives inside a stable frame, the stay, the gate, the clock and the costs rule. The amendment that failed is the one that promised a new frame and never commenced it. If the fourth amendment is enacted, its success will be measured by the standard the trail itself supplies: not by the architecture it announces, but by the notifications that follow it into the Gazette.

For the drafter, the trail reduces to four lessons, each bought with a provision. Commencement is legislation: a reform divided between enacted text and postponed operation is two reforms, and the second may never happen, so the operative centre of a scheme should never be left to a notification that a later government must affirmatively choose to issue. Applicability is substance: the date from which an amendment speaks determines rights in pending cases, and a clause that treats that determination as a technicality will be litigated as the merits, and may be struck as the merits were struck in 2019. Schedules are policy in miniature: a qualification schedule that reads as a national preference will be answered by the market it addresses, in this case by omission within two years and by a consultation draft that abandoned the device entirely. And retrospectivity is a rationale-bearing act: the fraud proviso survived its passage, on terms no court has disturbed, because it carried its justification on its face and in its debate, where Section 87, which carried none that answered the judgment it displaced, did not survive a season. A statute amended three times in six years, whose central institutional promise stands uncommenced and whose central applicability question was settled by a constitutional bench’s successor doctrines, is not a failed statute. It is a statute whose working text was chosen, provision by provision, by the interaction of the institutions that draft, commence, construe and strike, and the trail, rather than any single amendment, is the law’s true history.

Frequently Asked Questions

Q: What did the 2015 amendment do about arbitrator independence and timelines?

The amendment converted independence into a code and time into an obligation. Section 12 was recast to require written disclosure of circumstances likely to give rise to justifiable doubts about independence or impartiality, guided by the Fifth Schedule, and Section 12(5) made persons within the Seventh Schedule ineligible to act, subject to waiver only by express written agreement after disputes arose. On timelines, Section 29A required the award within twelve months of the tribunal entering upon the reference, with six months more by consent and court extension thereafter, while Section 29B offered a six-month fast track on documents alone.

Q: What did the 2019 arbitration amendment do about institutional appointment?

The amendment substituted Section 11 so that appointments on default would be made by arbitral institutions designated by the Supreme Court and the High Courts from among institutions graded by the Arbitration Council of India, removing the appointing function from the courts save where no graded institution was available. The design depended on the Council and the grading regime in Part IA, and that Part was never brought into force. The substitution therefore never commenced, and appointments continue to be made by the courts under Section 11 in the form the 2015 amendment gave it, confined to the existence of the arbitration agreement.

Q: What did the 2021 arbitration amendment do about fraud and stay?

The amendment inserted a further proviso into Section 36(3). Where the court is satisfied that a prima facie case is made out that the arbitration agreement or the contract which is the basis of the award, or the making of the award, was induced or effected by fraud or corruption, the court shall stay the award unconditionally until the Section 34 challenge is disposed of. The proviso is deemed to have been inserted with effect from 23 October 2015, and an explanation applies it to all court cases arising out of or in relation to arbitral proceedings, irrespective of when those proceedings commenced.

Q: Has the Arbitration Council of India been set up?

No. The Council was created on paper by Part IA, Sections 43A to 43M, inserted by the 2019 amendment, which designed its composition, its grading of arbitral institutions, its accreditation functions and its advisory role. The notification S.O. 3154(E) of 30 August 2019 commenced only Sections 1, 4 to 9, 11 to 13 and 15 of the amendment Act, and Section 10, containing Part IA, was not among them. No chairperson or member has been appointed under the Part, no institution has been graded by the Council, and no arbitrator has been accredited under Section 43J.

Q: What happened to the automatic stay on enforcement of an arbitral award?

The unamended Section 36 was read to suspend enforcement automatically once a Section 34 challenge was filed. The 2015 amendment ended that position: the award is enforceable on expiry of the time for challenge unless a separate stay application is granted, and a stay of a money award is imposed on conditions akin to a money decree, including deposit. Section 87 of the 2019 amendment attempted to revive the automatic position for arbitrations begun before 23 October 2015, and the Supreme Court struck Section 87 down in Hindustan Construction. A qualified exception for fraud cases arrived with the proviso of 2021.

Q: How did public policy as a ground narrow?

Before 2015, the public policy ground in Section 34 had been read, following Saw Pipes, to include patent illegality, which invited review of awards on points of law. The 2015 amendment confined the ground by explanation: an award conflicts with the public policy of India only where its making was induced or effected by fraud or corruption, where it contravenes the fundamental policy of Indian law, or where it conflicts with the basic notions of morality or justice, and the explanation bars merits review under that cover. Patent illegality survived as a separate ground in Section 34(2A), confined to arbitrations other than international commercial arbitration.

Q: What happened to the arbitrator qualification schedule?

The Eighth Schedule, inserted by the 2019 amendment, prescribed the qualifications, experience and norms for accreditation as an arbitrator under Section 43J, and its categories, framed around Indian professional and service backgrounds, had the effect of excluding foreign qualified practitioners. The schedule was criticised by the international arbitration community and by domestic practitioners as a closed-market measure in a statute competing to attract international work. It never came into force, because Part IA was never commenced, and the 2021 amendment omitted the schedule altogether, substituting Section 43J so that qualifications and norms would be specified by regulations.

Q: What is proposed in the latest arbitration reform?

The Draft Arbitration and Conciliation (Amendment) Bill 2024, released for public consultation by the Department of Legal Affairs on 18 October 2024 on the basis of the expert committee report chaired by Dr T. K. Viswanathan, proposes to rename the statute the Arbitration Act 1996 by removing conciliation, which the Mediation Act 2023 covers separately. It proposes statutory recognition of emergency arbitrators through a Section 9A, an appellate arbitral tribunal as an opt-in alternative to a Section 34 challenge, timelines for appointment and challenge, arbitration by audio-visual electronic means within the definition, and a restated Arbitration Council with altered composition.

Q: Why did Parliament amend the Arbitration Act in 2015?

Parliament acted on the 246th Report of the Law Commission of India, submitted in August 2014, which documented why the 1996 Act was failing its purpose. Court appointment applications had become a preliminary litigation tier argued on the merits. The public policy ground, as read in Saw Pipes, permitted rehearing of awards on points of law. Enforcement froze automatically when a challenge was filed, as National Aluminium had held, so award holders waited years beyond the award. Arbitrations had no statutory deadline and ran as long as the suits they replaced, and interested arbitrators faced no disclosure or disqualification regime. The amendment answered that catalogue.

Q: What did the 2015 amendment change in Section 8 on court referral to arbitration?

The amendment recast Section 8 so that a judicial authority, faced with an action in a matter that is the subject of an arbitration agreement, shall refer the parties to arbitration unless it finds prima facie that no valid arbitration agreement exists. The referral application must be made not later than the date of submitting the first statement on the substance of the dispute, and the authority is directed to act notwithstanding any judgment, decree or order of a superior court in the same matter. The change tightened the earlier discretion and aligned the referral gate with the prima facie standard applied at appointment under Section 11(6A).

Q: What did Section 26 of the 2015 Amendment Act provide, and what became of it?

Section 26 of the Amendment Act provided that nothing in the amendment Act shall apply to arbitral proceedings commenced before its commencement, unless the parties otherwise agreed, while applying the amendments to arbitral proceedings commenced on or after that date. The text divided the world by the date of the arbitration and said nothing express about court proceedings. High Courts divided on whether the amended Section 36 reached post-commencement Section 34 applications in pre-commencement arbitrations, until Board of Control for Cricket in India v Kochi Cricket Partners held that it did. Section 26 was deleted by the 2019 amendment, and that deletion was struck down with Section 87 in Hindustan Construction, so the Kochi Cricket construction governs.

Q: What did the Supreme Court hold in Board of Control for Cricket in India v Kochi Cricket Partners?

The Court held in 2018 that the 2015 amendment’s changes to court-facing provisions apply to court proceedings instituted on or after 23 October 2015 even where the underlying arbitration began earlier. Reading Section 26 of the amendment Act as addressing arbitral proceedings, the Court applied the amended Section 36 to all Section 34 applications filed after the commencement date, removing the automatic stay of enforcement for that entire class. The judgment treated the award holder’s right to execute as an accrued right and the removal of the stay as procedural, and it is the fixed point against which the 2019 amendment’s Section 87 was an attempted legislative reversal and the Hindustan Construction strike a judicial restoration.

Q: Which committee’s report led to the 2019 arbitration amendment, and what was its central proposal?

The High Level Committee to Review the Institutionalisation of Arbitration Mechanism in India, chaired by Justice B. N. Srikrishna and reporting on 30 July 2017, supplied the design. Where the Law Commission’s 246th Report had diagnosed court behaviour, the Srikrishna Committee diagnosed structure: Indian arbitration was ad hoc, institutions were weak, and appointments ran through courts or unilateral party nomination. Its central proposal was a statutory Arbitration Council of India to grade arbitral institutions, accredit arbitrators and promote institutional arbitration, with appointments routed through designated graded institutions. The 2019 amendment enacted that design in Part IA, which was never commenced, so the Committee’s central proposal remains law on paper and not law in operation.

Q: What did the 2019 amendment change in Section 34 on the material before the court?

The amendment altered Section 34(2)(a) so that a challenge is to be established on the basis of the record of the arbitral tribunal. The earlier language, which spoke of a party furnishing proof that a ground was made out, had left room for evidence beyond the tribunal’s record to be led before the court. The substituted wording confines the challenge to the material the tribunal had, which limits a party’s ability to improve its factual case at the setting-aside stage and keeps the court’s function a review of the award against its record rather than a rehearing. The provision was among those brought into force on 30 August 2019.

Q: How did the 2019 amendment change the twelve-month period for making an award?

The recast Section 29A moved the starting point of the twelve-month period from the date the tribunal entered upon the reference to the date of completion of pleadings under Section 23(4), and added Section 23(4) itself, requiring the statement of claim and defence to be completed within six months of the arbitrators receiving notice of their appointment. It also withdrew the mandatory timeline from international commercial arbitration, where the obligation became an endeavour to dispose of matters expeditiously. The recast recognised that the original trigger was too early in practice, since constitution formalities consumed the clock before substantive work began.

Q: What did the 2015 amendment provide on the fees payable to arbitrators?

The amendment addressed fees on two fronts. The Fourth Schedule set out a model fee for court-appointed arbitrators scaled to the sum in dispute, displacing per-hearing billing that rewarded adjournment, and Section 11(14) directed the court to have regard to that model when appointing. Separately, Section 29A gave the court, when extending the award timeline, a power to reduce the arbitrator’s fees where delay was attributable to the tribunal, and Section 31A required tribunals to weigh the conduct of the parties, including dilatory behaviour, when awarding costs. The schedule was a model and parties remained free to agree their own terms.

Q: What does the 2021 amendment’s retrospective reach mean for court proceedings begun before it?

The Act is deemed to have come into force on 4 November 2020, the date of the ordinance it replaced, while the fraud proviso is deemed to have been inserted into Section 36(3) with effect from 23 October 2015. The explanation to the proviso applies it to all court cases arising out of or in relation to arbitral proceedings, irrespective of whether the arbitral or court proceedings were commenced before or after the amendment. In plain terms, a party facing enforcement in an ongoing case may invoke the unconditional fraud stay for proceedings begun at any point since the 2015 settlement, and executions completed and stays refused in that window are subject to reopening where the prima facie threshold is met.

Q: Who is bound by the confidentiality duty the 2019 amendment created?

Section 42A, inserted by the 2019 amendment and brought into force on 30 August 2019, binds three classes: the arbitrator, the arbitral institution, and the parties to the arbitration agreement. The duty is not absolute. The section creates an exception for disclosure of the award where disclosure is necessary for its implementation and enforcement, so a party enforcing an award through a court is not barred from placing the award before that court. The provision was the statute’s first express confidentiality rule, since the 1996 Act as enacted had carried no general duty covering institutions and parties.

Q: Can a foreign national act as an arbitrator in an India-seated arbitration?

Yes. Nothing in the working law confines appointment to Indian nationals, and international commercial arbitration seated in India routinely involves tribunals of mixed nationality. The question gained currency from the Eighth Schedule inserted by the 2019 amendment, whose categories of qualification and experience, framed around Indian professional and service backgrounds, would in operation have excluded foreign qualified practitioners from accreditation, in tension with the statute’s ambition to attract international work to Indian seats. The schedule never came into force and was omitted by the 2021 amendment. The uncommenced accreditation scheme therefore never operated as a nationality barrier.

Q: What did the Constitution Bench decide on unilateral appointment clauses in 2024?

In Central Organisation for Railway Electrification v M/s ECI SPIC SMO MCML (JV), 2024 INSC 857, decided on 8 November 2024 by five judges, the Court held that equal treatment of parties at the appointment stage is a component of the validity of an arbitration agreement. The majority, speaking through Chandrachud CJI with Pardiwala and Misra JJ, held that clauses in public sector contracts permitting one party to appoint a sole arbitrator unilaterally, or to confine the counterparty’s choice to a panel curated by that party, fail the equality standard, drawing on the independence doctrine of the amended Section 12(5) and the Seventh Schedule. Roy and Narasimha JJ delivered separate opinions, and the operative directions were made prospective in the terms recorded at paragraph 170.7, expressed for three-member tribunals. …[truncated 9948 chars]