On September 1, 2026, a five-member bench of the National Company Law Tribunal stayed the order the country had spent a week doing arithmetic on. It held that no clear majority view existed, issued a notice to every party, said it would hear the matter afresh, and restrained Subhash Chandra Goenka from alienating any property, directly or indirectly.
Read that last direction again. Seven years after the Essel group’s defaults began and four years after this guarantee was invoked, the first enforceable restriction on this guarantor’s ability to move assets came from a tribunal, on its own motion, at an interim stage in a contested proceeding.
It did not come from a loan covenant. There was no loan covenant. That is the whole argument of this piece, delivered by the record itself.
On August 25, a third member of the tribunal approved a repayment plan setting aside Rs 6.25 crore for creditors and Rs 25 lakh towards the cost of the process, against admitted claims of roughly Rs 22,006.57 crore. Within hours the commentary had written itself. A 99.97 per cent haircut. Two and a half paise in the rupee.
Alongside it sat a second number. Insolvency and Bankruptcy Board of India (IBBI) data as of June 2026 shows about Rs 234.56 crore recovered across 64 concluded resolutions of personal guarantors since the chapter was notified in FY20, against total guarantor debt of roughly Rs 2.86 lakh crore as of December 2025.
Barely one per cent, down from the 2.16 per cent reported in early 2024.
The two were read as a single indictment of promoters. They are not. They are a receipt. One is issued for two decades of credit practice that treated a promoter guarantee as a signature rather than as security.
The other is issued for a chapter of the Insolvency and Bankruptcy Code that was drafted in a hurry, notified in fragments, and never given the safeguards its corporate counterpart takes for granted. As of this week, the first of those numbers is not even an operative order. The problem it pointed to has not moved an inch.
The matter arose from a 2022 application by Indiabulls Housing Finance, since renamed Sammaan Capital, invoking a personal guarantee furnished for a loan of about Rs 170 crore to Vivek Infracon.
The tribunal admitted the plea under Section 95 in April 2024. The repayment plan that followed went to a vote and secured 80.81 per cent of the voting share. Objecting creditors, led by LIC Housing Finance, held less than a fifth.
Then the bench split, and kept splitting.
Judicial Member Ashok Kumar Bhardwaj would have approved the plan only as against the creditors who voted for it, leaving dissenting banks and financial institutions free to pursue independent remedies. Technical Member Reena Sinha Puri rejected the plan outright, finding serious defects in the process followed by the resolution professional.
The difference went to a third member, Nilesh Sharma, under Section 419(5) of the Companies Act, 2013.
On August 25, he approved the plan under Section 114, excluding claims filed by Anil Kumar on behalf of 960 individuals and through Sunil Jain on behalf of 300, directed that their share be redistributed among the remaining eligible creditors, and held under Section 115 that the plan bound all creditors, dissenters included.
Three members. Three destinations. On August 31, the original bench recorded what was by then obvious: no majority view had emerged, because one member would bind nobody but the consenting, one would bind nobody at all, and one would bind everyone.
The President constituted a five-member bench — Justice (retd) Anupinder Singh Grewal, with Judicial Members Bachu Venkat Balaram Das and Mahendra Khandelwal and Technical Members Atul Chaturvedi and Ravindra Chaturvedi — which sat on September 1, and stayed the whole thing.
Three points of law are worth stating plainly, because they are still being misreported.
First, this is not the insolvency of Zee Entertainment, and it does not touch what any company owes anyone. It concerns what one individual owes on guarantees he signed.
Second, under Section 128 of the Contract Act, the liability of a surety is coextensive with that of the principal debtor, and resolving the surety does not extinguish the debt at source.
The Supreme Court settled the mirror image of that proposition in Lalit Kumar Jain in 2021. Rs 22,000 crore was never written off. Even at its high-water mark, the plan only redirected it.
Third, and this one is new. Section 419(5) is built for a binary disagreement: two members differ, a third breaks the tie. It has no answer when three members reach three genuinely different conclusions on the same record, because there is then no tie to break and no majority to reconstruct.
The provision was stretched to its limit in this case, and the limit gave way in six days. A tribunal that has to convene five members to discover what its own file says is not a tribunal behaving badly. It is a tribunal being asked to do statutory work with an instrument that was not designed for it — which, as it happens, is exactly what the banks did with the guarantee.
There is a technical caveat worth stating before the outrage. The denominator is the whole book; the numerator comes from a very small number of closures.
Around 5,186 applications have been filed to date, of which some 4,215 were filed by creditors under Sections 94 and 95. Sixty-four have produced a recovery. The ratio is therefore not measuring how badly resolved cases perform. It is measuring how few cases resolve at all.
That caveat does not rescue the number. It relocates the problem. A framework in which barely one per cent of filings reaches an outcome after six years is not being defeated by clever defendants. It is failing on its own terms. This month it failed in public, on the largest file it has ever handled.
The recovery outcome is decided long before a tribunal sees the file. It is decided at three points, and creditors control the first two.
At origination. For two decades, the promoter guarantee has been collected as a signature rather than constructed as security. In most sanction files I have reviewed there is a net-worth certificate, sometimes a chartered accountant’s letter, and nothing else.
No schedule of identified guarantor assets. No negative pledge over those assets. No obligation to report material disposals. No periodic refresh of the certificate. An unsecured, unmonitored, unrefreshed personal covenant is not credit enhancement. It is a comfort document, and comfort is not recoverable.
At enforcement. Guarantees are typically invoked years after the corporate account has soured, often only once the corporate resolution has closed and the shortfall has been quantified.
By then, the guarantor’s balance sheet has had a full cycle to reorganise itself through trusts, family transfers, offshore holdings and pledged-share enforcement, none of which breached any covenant because no covenant addressed them. Recovery from a guarantor is overwhelmingly a function of how early you move. Indian lenders move last.
On September 1, a tribunal had to supply, by interim order, the restraint on asset movement that a competently drafted facility agreement would have carried on day one.
At adjudication. Only here does the statute become the binding constraint, and the constraint is real. This is where the corporate and individual frameworks part company, and where I believe the reform conversation belongs.
One. There is no value floor. In a corporate resolution, Section 30(2)(b) guarantees a dissenting financial creditor at least what it would have received in liquidation.
That single provision converts a creditor vote into a vote constrained by an objective benchmark. Part III has no equivalent. Sections 111 and 114 require a three-fourths majority in value and little else. A plan can therefore be approved because enough creditors said yes, without any statutory requirement that the dissenters be shown to be no worse off than in bankruptcy.
The three-way split is the clearest possible demonstration. Three members of the same tribunal, applying the same chapter to the same record, reached three different answers to the question of whom a repayment plan should bind. That is not a judicial idiosyncrasy.
That is what happens when a statute supplies a voting threshold and no benchmark against which to test the outcome of the vote. Give three careful people a majority rule and no floor, and they will disagree about what the majority has actually bought.
Two. There is no related-party voting bar. The proviso to Section 21(2) keeps related parties out of the committee of creditors in corporate insolvency, for the obvious reason that a debtor should not be able to vote on his own settlement through intermediaries. Section 5(24A) defines what a related party means in relation to an individual. That definition was never carried into the voting architecture of Part III.
The record in this case shows why that matters. Claims of 960 individuals were routed through one person and 300 through another, and the third member found them fit to be excluded, with the tribunal noting lapses in their admission on the strength of verbal assurances. Set aside the merits of any particular allegation.
The structural point stands on its own: 1,260 claims entered a voting register through two intermediaries and had to be removed by a judicial finding at the plan-approval stage, because nothing in the chapter stopped them entering it. A voting threshold is only as robust as the register of persons entitled to vote, and Part III does not police the register.
Three. There is no independent valuation of the debtor’s estate. Corporate insolvency has registered valuers, prescribed methodologies and a fair-value-versus-liquidation-value exercise under the CIRP Regulations.
The personal insolvency framework leans heavily on what the debtor discloses to the resolution professional. In this matter, four figures for one man’s worth are on the public record across a decade, ranging from tens of thousands of crores certified to lenders in 2017 and 2018 down to about Rs 31.79 crore disclosed in 2024.
Some of that gap is the honest destruction of value that follows a leveraged group unwinding. Some of it may not be. Part III contains no forensic mechanism to find out.
The avoidance provisions in Sections 164 to 167 exist, but they depend on someone with resources and standing to run them, and the look-back window for undervalued transactions is a modest two years, which is a short memory for assets that can be moved within a family in an afternoon.
A technical member has now recorded serious defects in the process by which the estate and the claims were assembled. That finding is not an accusation against an individual professional. It is the predictable result of asking one professional to verify an estate and a register that the statute equips nobody to verify.
There is a second audience for these proceedings, and it is not the tribunal. Lenders who priced credit on the strength of a promoter’s name and did nothing to track that name’s balance sheet have been discovering the difference in tribunal after tribunal.
The Chandra file is simply the most legible instance so far, and the order of September 1, makes the point with a bluntness no commentary could improve on: the covenant these creditors never took has been supplied to them by a bench, seven years late, as an interim measure in a case that has not yet been decided.
The corrective is contractual and it is available today. Refresh guarantor net worth annually and have it audited. Take an identified asset schedule and a negative pledge over it. Cross-verify guarantor disclosures against filings made to other regulators and authorities. Restrict the transfer of identified guarantor assets.
Retain a right of inspection that survives default. Build guarantee invocation into the early-warning framework rather than leaving it to the recovery department’s last resort. Every one of these sits within existing contract law. None of it needs the regulator’s permission.
If you force me to apportion, I would put the larger share on creditors and the more urgent fix on the statute. Banks wrote a weak instrument. Parliament wrote a chapter that has no capacity to repair a weak instrument.
A tribunal applying Sections 111, 114 and 115 to a guarantee that was never properly secured, never monitored and invoked five years late will produce a small number, and it will have applied the law correctly in doing so.
I would also resist the framing of promoters getting away. Most guarantors are not executing a scheme. They are responding rationally to a system whose expected cost of enforcement they can estimate with reasonable accuracy: four to six years, high legal spending, and a settlement negotiated from a position of considerable leverage.
This week has added a data point to that estimate, and it does not favour the creditor. When the expected cost of resisting is lower than the cost of paying, resistance is the commercially correct answer. That is a design defect, not a character defect, and treating it as a morality story is precisely why we keep failing to fix it.
It is fair to record that the regulator has not been idle. The IBBI’s discussion paper of July 2026 proposes tightening the valuation framework by requiring creditors’ approval for the appointment of valuers and keeping valuation reports confidential until plans are received.
It also proposes operationalising the withdrawal of interim moratorium protection for personal guarantors following the Amendment Act, 2026. Those two changes go to the heart of the enforcement and adjudication failures. The interim moratorium, in particular, had become a tactical instrument: filing an application froze parallel remedies, and delay favoured the debtor.
The Insolvency and Bankruptcy Board and the Ministry of Corporate Affairs do not need a new statute. They need Part III to be brought up to the standard Part II has held for nine years. Five amendments would do most of the work.
Introduce a comparator test for repayment plans, requiring that a dissenting creditor receive no less than its share of the bankruptcy estate, and that the adjudicating authority record a finding on it.
Extend the related-party disqualification to voting on repayment plans, using the definition Parliament has already enacted, and require the resolution professional to certify the voting register against it.
Mandate an independent valuation of the debtor’s estate by a registered valuer, with a reasoned reconciliation wherever disclosed worth departs materially from net worth certified to lenders or to any public authority.
Lengthen the avoidance look-back for individuals, where assets are far easier to reorganise within a family than within a balance sheet. And make interim protection of the estate the default on admission rather than a discretionary order arrived at years later, so that a bench never again has to invent a negative pledge in the middle of a contested hearing.
A sixth, procedural point now recommends itself. Section 419(5) needs an express rule for the case where members reach more than two distinct conclusions. The present position produced a fortnight of confident national commentary about an order that turned out not to exist. That is avoidable by drafting.
Part III was written to give ordinary insolvent individuals a route back to solvency, and that purpose is legitimate. The framework is now carrying two entirely different populations: the salaried debtor with a defaulted loan, and the promoter with guarantees running into thousands of crores.
Their risk profiles, their access to advice and their capacity to structure assets are not comparable. Tightening the chapter uniformly will punish the first group in order to reach the second.
The honest reform is to separate them, with a distinct high-value guarantor track carrying stricter disclosure, valuation and avoidance rules.
The most damaging reading of this fortnight would be that the tribunal disgraced itself. It did not. A third member applied Section 114 to a live question, the original bench found his conclusion did not yield a majority, the President convened a larger bench within a day, and that bench stayed the order and protected the estate before hearing anyone.
Six days from opinion to correction, with the debtor’s assets frozen in the interim, is an institution working. Very few forums in this country would have moved that quickly to check themselves.
But institutional health is not the same as statutory health, and it would be a mistake to let the first be offered as evidence of the second.
Five members were needed to determine what three members had decided. The estate had to be protected by judicial improvisation because no contract protected it. The register of voters had to be cleaned by a judicial finding because no provision kept it clean.
Each of those is a bench compensating, in real time, for something the chapter should have supplied in advance.
The uncomfortable truth is harder to shout about than a percentage. One chapter of one statute is being asked to solve two unrelated problems, on the strength of an instrument the banking system never bothered to secure.
Until Parliament acknowledges that, we will keep producing proceedings that are legally sound and publicly indefensible, we will keep quoting the one per cent as evidence of villainy when it is really evidence of drafting, and we will keep mistaking the second for the first.
Two and a half paise in the rupee was never the scandal. This week it is not even the order. What remains, and what will still be there when the five-member bench finishes, is a signature that no one secured and a chapter that cannot make up for it.
(The views in the article are personal and do not reflect those of the organisation. The above article is authored by Akshat Khetan, the Founder of AU Corporate Advisory and Legal Services.)