The IBC implementation record is usually argued from a single number, and the number is almost always the haircut. A commentator takes the admitted claims of creditors in cases that produced a resolution plan, takes what the plan paid, divides the second by the first, and announces that the Insolvency and Bankruptcy Code, 2016 has returned roughly a third of what was owed and destroyed the rest. The arithmetic is not wrong. The inference is, because the admitted claim measures what a failed company’s creditors were promised across years of lending against deteriorating security, while the plan’s payment measures what the company’s assets could fetch in a market sale conducted under time pressure after the failure had already happened. Between those two figures sits the entire history of the company’s decline, almost none of which occurred inside the process being judged. An implementation record cannot be read off one ratio, and the regulator that administers the Code, the Insolvency and Bankruptcy Board of India, does not publish one ratio. It publishes several, against different denominators, and the discipline of keeping them apart is the whole of the method this article applies.

That discipline produces a verdict in three parts, and the parts point in different directions. Measured against what liquidation of the same companies would have realised, the Code’s resolution process has performed strongly and consistently, returning substantially more than the break-up value of the assets in case after case. Measured against the clock the statute itself set, the record is poor and has deteriorated, with average completion times running at close to double the outer limit the law prescribes and rising rather than falling across successive reporting periods. Measured by the effect the Code was most deliberately designed to produce, the change in debtor behaviour before a case is admitted at all, the record is very strong, with tens of thousands of applications withdrawn before admission against underlying defaults measured in lakh of crore of rupees. Each of those findings is documented in the Board’s own published reporting, attributed to a named source and a stated period in what follows. None of them can be merged into a single percentage without destroying the information the others carry, and the habit of merging them is responsible for most of the confident nonsense written about this statute in both directions.

What the Code promised, and the objective the record must be judged against

The first methodological error in most assessments of the Code is a mistake about its purpose. The Insolvency and Bankruptcy Code is habitually described as a recovery law, a machine for getting banks their money back, and then graded on the money it gets back. The Bankruptcy Law Reforms Committee chaired by T K Viswanathan, whose report supplied the Code’s design, framed the objective differently: the law was to resolve the insolvency of a firm in a manner that protected the value of the enterprise, keeping a viable business alive under control of those with money at stake and closing an unviable one swiftly, so that capital and labour locked in failure could move to productive use. Creditor recovery, in that design, is a consequence of preserving or realising value well, not the objective itself. The statute’s own long title speaks of maximising the value of assets and promoting entrepreneurship and the availability of credit, in that order, and the structure of the process, which removes the defaulting management on admission and hands the company to a committee of its financial creditors, is a control mechanism dressed as a recovery mechanism. The full architecture of that process, its sections, thresholds and case law, is set out in the pillar guide to the Insolvency and Bankruptcy Code 2016, and this article takes that structure as given and asks only what the evidence shows about how it has worked.

Judged against its own objective, the Code invites three questions, and only three, and they are the questions the Board’s reporting architecture is built to answer. The first is whether resolution realises more than liquidation of the same assets would have realised, because that is the comparison that isolates the process’s contribution from the company’s prior decline. The second is whether the process completes within a time that preserves the value it is trying to distribute, because a resolution mechanism that takes years consumes in delay much of what it saves in value. The third is whether the existence of the process changes behaviour outside it, because a credible threat of losing control is worth more to a credit system than any single case outcome. The haircut against admitted claims answers none of these questions. It compares the outcome not with the alternative the law faced but with the promises the borrower made before default, and no insolvency system in any jurisdiction, measured that way, escapes a large apparent failure, because the denominator is the debt of companies selected precisely by their inability to pay it.

What was the Insolvency and Bankruptcy Code actually designed to achieve?

The Code was designed to resolve a failing firm quickly and at the highest value its assets could command, by transferring control from the defaulting management to the creditors on admission, and to make that threat credible enough that debtors would settle rather than lose the company. Recovery for creditors follows from that design; it is not the design.

The committee that designed the Code was explicit that the decision on viability belonged with the creditors who had money at stake, not with a tribunal reconstructing the business, and that speed was not an administrative nicety but the substance of the reform, because a distressed firm’s value drains away while forums deliberate. That is why the process was given statutory timelines measured in days, why the moratorium freezes competing enforcement the moment a case is admitted, and why the committee of creditors’ commercial decision on a plan was placed substantially beyond judicial second-guessing. Every one of those design choices generates a measurable prediction. If control transfer works, debtors settle before admission in large numbers. If creditor control works, resolution beats liquidation value. If the timeline discipline works, cases conclude near the statutory limit. The implementation record is simply the set of answers the published data gives to those three predictions, and it gives a different answer to each, which is what makes a one-number verdict impossible for an honest reader.

One element of the design can be tested directly in the filing data, and it passes. Creditor-in-control presupposes that creditors, and in particular the financial creditors who hold the committee votes, would actually use the power the statute gave them, and the initiation pattern shows that they did: the process was taken up at scale by banks and financial institutions from its first years, including against the largest and most politically connected defaulters in the economy, a class of borrower that the older regime’s negotiations had conspicuously failed to dislodge. The moratorium’s credibility can be tested the same way, in what did not happen: the parallel recovery actions, attachments and enforcement proceedings that the moratorium was designed to freeze did not continue as a rival system alongside the process, because the courts held the freeze to its terms across the run of cases. A design can fail in its timelines and still succeed in its allocation of power, and the record keeps the two findings separate because the evidence for each comes from a different place: the allocation from the fact and pattern of the filings themselves, the timeline from the Board’s averages. The sections that follow are organised on that separation, value and behaviour first in the weighing, calendar never excused by either.

The record in phases: why no single period describes the Code

The cumulative figures conceal a sequence, and the sequence matters because the Code that produced the early data was not operating in the conditions that produced the later data. In the first phase, from commencement of the corporate provisions in December 2016 through 2018, the system was being built while it ran: benches of the National Company Law Tribunal were constituted, the first cohorts of insolvency professionals were registered, the regulations were written in response to problems as they surfaced, and the caseload was swollen at the outset by direction rather than by ordinary filing. In June 2017 the Reserve Bank directed proceedings against twelve large accounts that together represented about a quarter of the banking system’s gross non-performing assets, and those references, Essar Steel and Bhushan Steel among them, dominated the early statistics by value and taught the system its procedures in public. Outcomes from that phase measure a start-up institution processing the largest and most contested failures in the economy with a statute whose leading questions, on admission, on the treatment of homebuyers, on the promoter bar, had not yet been answered by any appellate court.

The second phase, running through 2019, was the phase of legal consolidation and legislative reaction. The courts supplied the architecture’s missing answers in sequence: the constitutionality of the Code was upheld, the committee of creditors’ commercial wisdom was placed beyond merits review, and the distribution discretion that produced the operational creditor grievance was confirmed. Parliament legislated in the same period, inserting the Section 12A withdrawal route in 2018 and the 330-day outer limit in 2019, each amendment a response to a behaviour the data had revealed, gaming of the calendar in one case and abandonment of settled cases in the other. The third phase was an interruption no assessment may treat as ordinary data. In March 2020 the minimum default threshold for initiation was raised from one lakh rupees to one crore, and Section 10A then barred initiation altogether for defaults arising during the pandemic period, a suspension that ran for defaults occurring up to March 2021 and permanently immunised that cohort of defaults from the process. Admissions fell sharply in the suspension year, and the cases that did proceed were disproportionately the legacy stock already inside. Any year-by-year comparison that crosses 2020 without flagging the break is comparing two different legal regimes wearing one name.

The fourth phase, from the pandemic suspension’s expiry onwards, is the mature record, and it is the phase on which the Code will ultimately be judged. Its characteristics are stability in the value ratios and deterioration in the timelines: realisation settled into its narrow band of roughly a third of claims and roughly one hundred and seventy per cent of liquidation value, the resolution-to-liquidation balance improved as the inherited caseload cleared, and the average completion time lengthened year after year. That combination is the single most informative pattern in the whole record. It rules out the comforting theory that the system’s weaknesses were transitional, the friction of a new institution learning its work, because a learning institution’s timelines shorten as its ratios stabilise. What the mature phase describes instead is a process whose value machinery works and whose calendar does not, operating at a scale its tribunal infrastructure was never expanded to match. The sections that follow take the measures in turn, but the reader should hold the phases in mind throughout: an average struck across all four phases describes none of them exactly, and the Code’s defenders and critics each prefer the phase whose data suits them.

The reporting architecture: who counts, and what the counts contain

Almost everything the public knows about the Code’s operation comes from one source: the quarterly newsletters published by the Insolvency and Bankruptcy Board of India, which tabulate the corporate insolvency resolution processes admitted before the National Company Law Tribunal and report their outcomes in fixed categories. A process is counted when it is admitted. It leaves the ongoing stock in one of a small number of ways: approval of a resolution plan, an order for liquidation, withdrawal under Section 12A of the Code, or closure on appeal, review or settlement. The newsletters also carry the money figures, reporting for cases that yielded resolution plans the admitted claims, the fair value and liquidation value of the assets as assessed on entry, and the amount the approved plans are to realise. Much of the confusion in public argument is a failure to notice that these are three different monetary magnitudes attached to the same cases, and that a percentage is meaningless until its denominator is named. The way this series has handled enforcement statistics in another field, in the implementation record of the anti-terror statutes, turns on the same rule: a figure quoted without its base and its period is not evidence, however often it is repeated.

Two features of the architecture deserve emphasis because each is a standing trap. The first is the gap between a resolution plan’s approval and the realisation it promises. The newsletters record the realisable value under approved plans, which is the amount the successful applicant has committed to pay, in the main over time and subject to the plan’s implementation; approval is the legal event the statistics count, and the money follows under monitoring afterwards. Treating an approval as cash already recovered overstates the record, and the Board’s figures should be read as committed realisation in the cases counted, which is how this article reads them. The second trap is the parallel gap inside liquidation: a liquidation order is the beginning of a separate process of asset sale and distribution that runs for years, so a count of liquidation orders in a period is not a count of completed liquidations, and the recovery figures for the two populations differ by an order of magnitude. The newsletters keep the categories distinct. Commentary routinely does not, and then marvels at contradictions the underlying tables never contained.

The reporting also carries a quieter caveat that any careful reader must hold: the Board’s tables are compiled from reports filed by insolvency professionals, and cases are occasionally reported late, so cumulative totals are revised between editions and a figure quoted from one quarter’s newsletter may differ slightly from the same cumulative figure in the next. That is not a scandal; it is the ordinary behaviour of a reporting system fed by thousands of practitioners. It does mean that precision to a decimal point, quoted without the edition and period it came from, is false precision. The figures in this article are therefore each tied to the publication that carried them and the date to which they run, and where two official cuts of the record give different values for overlapping magnitudes, both are stated with their bases rather than averaged into a number no source ever published. Readers who download the Board’s newsletter data annexes to test any of this can open the Board’s presentation annexes with a PowerPoint viewer and check the category totals against the argument directly, which is the use the Board publishes them for.

Is a case counted as resolved when the plan is approved or when the money is paid?

The Board’s statistics count the legal event: a case enters the resolution column when the tribunal approves the plan, and the amount recorded is the realisable value the plan commits to pay. The money itself is disbursed under the plan over its implementation period, so the published realisation figures are commitments recorded at approval, not cash verified in creditors’ accounts.

That convention is defensible, because approval is the moment the process the Code controls comes to an end and the moment the comparison with liquidation value is struck, but it places a duty on the reader. When a plan later fails in implementation, the case does not migrate back out of the resolution column in the headline tables, and the small number of approved plans that collapse into liquidation afterwards must be looked for in the liquidation process tables rather than in the resolution totals. The honest summary of the realisation evidence is therefore that approved plans commit sums well above liquidation value in the cases counted, that the overwhelming pattern is of plans being implemented because the applicant’s money is at stake from approval, and that the Board’s headline ratios describe the value of what was promised at the process’s end rather than an audited account of what every creditor ultimately banked.

The Board is not the only counter, and the subsidiary counters matter because each counts for a different purpose and therefore cuts the record differently. The Ministry of Corporate Affairs periodically places consolidated statements before the public and before Parliament, and those statements, being the government’s account of its own statute, tend to lead with the rescue count and the liquidation value multiple. The Reserve Bank counts for financial stability purposes, in its Report on Trend and Progress of Banking and in its Bulletin analyses, and its cuts attend to what the banking system recovered and through which channels, a creditor-side view in which settlements and channel shares loom larger than case counts. The Economic Survey, the finance ministry’s annual assessment, republishes the Board’s tables with the government’s interpretation attached, and has been the source of the most quoted timeline and pre-admission settlement figures. The rating agencies, ICRA and CARE among them, cut the data by financial year and by their own case tracking, which is why their haircut figures, computed on the flow of plans approved within a single year, can deepen sharply in a year when the cumulative ratios barely move. None of these counters is falsifying anything. Each is answering its own institution’s question, and the reader who knows which question a figure answers is most of the way to knowing whether it can bear the weight an argument loads onto it. The rule for what follows is that figures are used only for the question their source was asking, and are labelled with that source and its period every time.

One further distinction runs through all of the counters and is worth isolating before the numbers begin, because it converts many apparent contradictions into bookkeeping. The Board’s outcome categories are stages, not verdicts, and a single company passes through several of them in sequence. Admission is a stage. A liquidation order is a stage, after which the liquidation process itself runs for years and is counted in a separate set of tables with its own averages. Even dissolution, the formal striking of the company from the register at the very end, lags the economic events by years. A snapshot of the system at any date therefore shows a pyramid: a large stock of cases in the early stages, a smaller number that have reached an outcome stage, and a small residue that have traversed every stage to legal extinction. Arguments that compare the count at one stage with the count at another, liquidation orders against completed liquidations, applications filed against plans approved, without adjusting for the pipeline’s shape, are not measuring performance at all. They are measuring the age of the system, and they will flatter or condemn it according to nothing more than how long it has been running.

The system’s geography belongs in the architecture as well, because “the tribunal system” in the aggregate is, in load terms, a small number of courtrooms. Jurisdiction follows the corporate debtor’s registered office, and Indian incorporation is concentrated, so the Board’s bench-wise picture shows the caseload falling disproportionately on a few benches of the National Company Law Tribunal, Mumbai and Delhi foremost among them, with the long tail of benches carrying comparatively few cases each. The concentration explains a paradox in the capacity debate: sanctioning more benches nationally does little for the queue that matters, because the queue is where the companies are, and it also means the national averages are substantially the record of two or three benches’ listing practices, member strength and bar. It has a further consequence for the data’s texture. Where a case is filed affects its timeline in ways no national ratio discloses, and the parties know it, which is one reason registered-office questions surface inside admission litigation that presents itself as being about debt and default. A reader of the aggregate record should therefore hear every national average as a weighted average of local systems of unequal load and unequal speed, dominated by the busiest few.

What gets counted also depends on how default is proved, and the proof regime quietly shapes the population. Admission turns on evidence of debt and default, and the Code designed the information utility to be that evidence’s home: a record of default held by the utility is meant to settle the question, sparing the tribunal a trial of the debt. Bank debt, originated in documented facilities and increasingly reported to the utility, fits that design. Trade debt, arising from invoices, purchase orders and contested deliveries, mostly does not, which is one structural reason operational creditor applications attract heavier contest at the admission stage and financial creditor applications move through it faster: the two classes arrive carrying different qualities of proof, and the tribunal’s task differs accordingly. The counted record inherits this asymmetry. Cases are admitted, and so enter every statistic in this article, partly in proportion to how well documented the underlying credit was, and the insolvency population is therefore tilted toward formal, bank-intermediated finance and away from the informal credit on which much of Indian commerce runs. That tilt is not a defect of the statistics; it is a fact about the economy the statute can reach, and an implementation record is the right place to state it, because reforms aimed at the uncounted economy of undocumented credit will never show their effects in the Board’s tables at all.

The funnel shape deserves one more paragraph, because the top of the funnel is where most filings end. For every case admitted, several applications leave the system before admission, by withdrawal, by settlement, by rejection or by the applicant’s loss of interest, and the pre-admission settlement figures already cited, tens of thousands of applications involving more than ten lakh crore of rupees of underlying default, measure the largest single exit from the funnel’s top. An assessment that begins counting at admission therefore starts after the system’s most common interaction with debtors has already concluded, and it will systematically understate the statute’s behavioural reach while describing its adjudicated caseload accurately. The converse error also exists: arguments for the Code that add the pre-admission settlements to the resolutions and announce a combined triumph are mixing a counted outcome, plans approved by tribunals, with an uncounted one, settlements whose terms, completeness and durability no public record verifies. The funnel disciplines both errors. Its stages, in descending order of size, are applications filed, applications surviving to admission, processes reaching an outcome, processes reaching resolution, and resolutions implemented and performing, and every claim about the record can be located at one of those stages and judged against the stage’s own evidence. Much of the public quarrel about the Code is two speakers citing different stages of one funnel as if they were rival descriptions of the same stage.

The four ratios

The findable artifact of this article is a table of the four measures by which the Code’s performance is quoted, what each one actually measures, and the specific error committed when each is quoted alone. The first three are the Board’s own ratios for cases yielding resolution plans. The fourth is not a ratio within cases at all but a quantum outside them, and it is included because it captures the effect the Code’s designers valued most and which no within-process percentage can see.

Measure	What it measures	What it omits	Reported by	The error when quoted alone
Realisation as a share of admitted claims	What approved plans pay against everything creditors were owed in those cases	That the claims accumulated during the company’s decline, against assets already worth far less	IBBI quarterly newsletters	Treats the borrower’s broken promises as the law’s failure, and grades every insolvency system in the world as a failure
Realisation as a share of liquidation value	Whether resolution beat the break-up alternative for the same assets	That liquidation value is a low benchmark, set on assets already at salvage level	IBBI quarterly newsletters	Converts a comparison against a distressed floor into proof of success, and hides the timeline failure entirely
Realisation as a share of fair value	How much of the going-concern worth assessed on entry the plan captured	That fair value is an estimate struck at admission, available only for the cases where it was assessed	IBBI quarterly newsletters	Presents an appraiser’s estimate as a fact of nature and ignores that the case population differs from the other two ratios
Pre-admission settlement	The debt resolved because the process existed, before any case was admitted	Everything that happens inside the process; it measures the threat, not the throughput	IBBI statements; Economic Survey	Invisible in every recovery ratio, so a debate conducted in ratios alone deletes the Code’s largest measured effect

Read together, the four measures give the implementation record its actual shape. The first ratio, against admitted claims, has run at roughly a third across successive reporting periods and is the source of the haircut headlines. The second, against liquidation value, has run at well over one and a half times, in most periods around one hundred and seventy per cent. The third, against fair value, has run close to ninety per cent, a little below the going-concern value assessed when the cases entered. The fourth measure is denominated in applications and rupees of default rather than percentages: tens of thousands of applications withdrawn before admission, against underlying defaults that the Economic Survey put at Rs 10.2 lakh crore through March 2024. A debate that quotes only the first of these is not an assessment of a statute. It is an assessment of how much value Indian companies had already lost by the time anyone invoked the statute, which is a different and older failure.

Two of the four denominators are not facts of nature but professional estimates, and the estimates carry their own implementation history. Fair value and liquidation value are produced by registered valuers appointed within the process, two valuers in the standard case with a third engaged where the first two estimates diverge significantly, so the denominators against which the system’s success is scored are themselves judgments formed under the pressure of the case, by professionals whose young regulatory regime was still building its standards while the largest cases in the country’s history were being valued. The estimates are not beyond question, and the Board’s own data shows their coverage improving over time: the Economic Survey noted fair value estimates available for 964 of its 1,068 resolution cases through September 2024, which means the fair value ratio is computed on a subset from which the earliest, least documented cases are missing. A careful reader treats the liquidation value multiple as the sturdiest of the ratios, because liquidation value is anchored in identifiable assets, treats the fair value ratio as informative but softer, and treats both as estimates made inside the process rather than as market facts discovered outside it. That caution does not rescue the claims ratio’s critics or its defenders; it simply records that the measuring instruments were built by the same young system as the results they measure, and that some part of the ratios’ stability across periods reflects valuers learning the conventions of their new profession as much as companies performing alike.

The realisation record in its three denominators

Begin with the most complete official cut, the Ministry of Corporate Affairs’ statement of the position through 31 December 2024, drawn from the Board’s data and placed before the public in April 2025. By that date, 8,175 corporate insolvency resolution processes had been initiated since the corporate provisions commenced in December 2016, and 1,119 of them had yielded approved resolution plans. The realisable value under those plans was Rs 3.58 lakh crore, which the ministry stated as 162.79 per cent of the liquidation value of the same companies’ assets and 87.58 per cent of their fair value. Read that sentence slowly, because it contains the whole controversy in miniature. The plans paid, in aggregate, a sum more than one and three-fifths times what breaking the companies up would have fetched, and a little under seven-eighths of what registered valuers had assessed the businesses to be worth as going concerns when the cases were admitted. Whatever else is said about the Code, a process that consistently converts salvage-value assets into going-concern prices at that multiple is doing the specific thing it was built to do, and it is doing it in the cases where resolution is possible at all.

The missing denominator in the ministry’s statement is admitted claims, and it is supplied by other cuts of the same record, which must be kept separate because they cover different periods and slightly different case sets. A study by the Indian Institute of Management Ahmedabad, published through the Board in August 2023, computed realisation at 32 per cent of admitted claims and 168 per cent of liquidation value for the resolved cases it examined. An IBBI statement of the position through March 2025, made by the Board’s executive director Jitesh John, recorded 1,194 processes ending in resolution, realising Rs 3.89 lakh crore, at about 32 per cent of the total claims, about 170 per cent of liquidation value and 93.36 per cent of fair value. The Economic Survey of 2024-25, reporting the Board’s figures through September 2024, put the same magnitudes at 161 per cent of liquidation value and 86.1 per cent of fair value across the 964 cases in which fair value had been estimated, and expressed the claims comparison from the other direction, as an average haircut of about 69 per cent against admitted claims and about 14 per cent against fair value. These are not four rival truths. They are the same stable pattern photographed at different dates through slightly different windows: roughly a third of admitted claims, roughly one and two-thirds to one and seven-tenths times liquidation value, and somewhere between eighty-six and ninety-three per cent of fair value, the fair value figure moving with the subset of cases in which valuers’ estimates are available.

Two refinements protect the reader against misuse of even this careful statement. The first is that a fiscal-year cut can look far worse than the cumulative cut without either being wrong. The rating agency ICRA, analysing the plans approved within single financial years, reported the average haircut deepening to 73 per cent in the year 2023-24 from 64 per cent in the preceding year. An annual flow is dominated by whichever large cases happen to conclude in that year, and a year heavy with asset-light service companies will show haircuts the cumulative stock of steel and infrastructure resolutions conceals. The second refinement concerns who the resolved population is. In a substantial minority of the cases that produced plans, the company was already dead in commercial terms when it entered: the ministry’s December 2024 statement recorded that around 39 per cent of the processes yielding resolution plans had previously been before the Board for Industrial and Financial Reconstruction or were defunct. The Board’s March 2025 statement separated those cases out and reported that creditors in them realised 18.96 per cent of admitted claims and 150.33 per cent of liquidation value.

One more qualification attaches to every realisation figure in this section, and it concerns time rather than denominators. A resolution plan’s value is a schedule of payments, not a payment: plans typically combine an upfront component with deferred instalments spread over years, funded from the revived company’s cash flows or the acquirer’s future resources, and they may include instruments whose value depends on performance after approval. The Board’s headline figures record the value of plans as approved, which means the realisation ratios are, strictly, realisable-value ratios, statements of what creditors are scheduled to receive if the plans perform. Two consequences follow for the reader. The first is that the ratios are not discounted: a rupee promised in year five is counted with a rupee paid at approval, so the economic recovery is somewhat smaller than the published ratios state, by an amount no official series computes. The second is that the ratios carry the acquirer’s performance risk, which the plan-failure evidence later in this article shows is not theoretical. None of this makes the figures dishonest, since the same convention applies across periods and the comparison with liquidation value, itself an estimate of a hypothetical sale, remains meaningful. It does mean that the value record, like the timeline record, is stated on the system’s own conventions, and that the distance between a plan approved and a creditor paid is a quantity the published record gestures at rather than measures, one more reason the verdict at the end of this article is framed on the pattern of the evidence rather than on any single triumphant or damning number. Even the Code’s successes, that is, include a large block of revivals of companies the older regime had already written off, and their low claims recovery drags the aggregate ratio while their high liquidation-value multiple confirms the process added value to estates that had none left to lose.

The IIM Ahmedabad study adds the one piece of evidence the Board’s tables cannot supply, which is what happened to the companies afterwards. Following the resolved firms into their post-resolution accounts, the study found sales growth of 76 per cent in the years after resolution, a rise of 50 per cent in employee expenses, and the resolved firms reaching operational break-even by the third year. That evidence matters to the verdict because the Code’s objective was the continuation of viable enterprise, not the maximisation of a distribution. A resolution that pays creditors a third of their claims and returns a functioning, employing, investing company to the economy has succeeded on the statute’s own terms in a way a liquidation paying the same third never could, and the post-resolution trajectory is the only part of the record that tests that claim directly. It is also the part of the record with the smallest evidence base, one institutional study rather than a statistical series, and it should be weighted accordingly: corroboration of the design’s premise, not a second set of headline ratios.

The large-case distortion: why the aggregates swing

Running through every aggregate in this field is a distortion that a careful reader must correct for mentally, which is the weight of a small number of very large cases. The Indian insolvency caseload consists of thousands of small and mid-sized companies whose claims are measured in crore or tens of crore, and a few dozen giants whose claims are measured in tens of thousands of crore. The giants dominate every value-weighted ratio. When Essar Steel, carrying admitted claims in the region of fifty thousand crore of rupees, resolved under a plan with a reported value of about Rs 42,000 crore, that single case moved the cumulative realisation ratios for the entire system by visible amounts, and the strong prices the large steel cases commanded during an upswing in the steel cycle, Bhushan Steel passing to Tata Steel in 2018 for a reported consideration of about Rs 35,200 crore among them, flattered the system’s early value record in ways that had as much to do with commodity prices and strategic buyers as with institutional design. Conversely, a period in which the concluding cases are disproportionately asset-light, in services, contracting or real estate, will show haircuts that no procedural reform could have prevented, because there was never much asset value to find.

This is why the annual flow figures and the cumulative figures must be read as answers to different questions, and why the rating agencies’ year-by-year haircuts, which deepened to 73 per cent in the plans approved during 2023-24 on ICRA’s analysis, do not contradict the cumulative third-of-claims figure even though the two are constantly quoted against each other. The cumulative stock answers the question of what the system has done with everything it has been given, giants included. The annual flow answers the question of what the system did with the particular mixture of cases that happened to conclude in that year, and in a field where one steel plant outweighs a thousand trading companies, the mixture is the message. A serious assessment uses the cumulative figures for the verdict on the institution and the annual figures only for spotting which way the caseload’s composition is drifting, and treats any argument that swaps the two, quoting a bad year as the system’s record or the cumulative stock as the year’s performance, as an error of method before it is anything else. The distortion also explains the persistence of anecdote in the debate: every participant’s experience of the Code is of particular cases, and the cases a banker, a promoter or a supplier personally met are a sample drawn from the mixture, not from the record.

The timeline record: the Code’s clearest failure

On time, the record contradicts the statute to its face. The process was designed to conclude in 180 days, extendable by 90, and the Insolvency and Bankruptcy Code (Amendment) Act of 2019 then wrote an outer limit of 330 days into Section 12, expressed to include the time taken in legal proceedings, precisely because litigation had become the standard instrument for consuming the clock. Within months, in the Essar Steel litigation, the Supreme Court read that limit down: in Committee of Creditors of Essar Steel India Limited v Satish Kumar Gupta, decided in November 2019, the Court struck the word “mandatorily” from the amended provision and held that the adjudicating authority could extend beyond 330 days in exceptional cases, so the outer limit operates as a strong direction rather than an absolute bar. The reasoning was pragmatic, that a process on the verge of conclusion should not be destroyed by a calendar line, and its practical consequence was to return the timeline to the custody of the tribunals, where every interested party with a grievance could spend it. The case law is Article 55’s territory; its footprint in the timeline statistics is this article’s.

Why did the Supreme Court refuse to treat the 330-day limit as absolute?

In the Essar Steel judgment the Court struck down the word “mandatorily” in the 2019 amendment, holding that an insolvency process on the verge of bearing fruit should not be forced into liquidation by a rigid deadline. The limit survives as the governing standard, but the adjudicating authority may exceed it in exceptional circumstances.

The measured consequences are documented across successive official publications, and they describe deterioration, not teething trouble. The Economic Survey of 2024-25, reporting the Board’s data through September 2024, recorded that the 1,068 processes that had yielded resolution plans took an average of 582 days to conclude, and that figure expressly excludes the periods the adjudicating authority excluded from the count in individual cases, which averaged a further 116 days. The exclusion mechanism itself is instructive: the statute permits the adjudicating authority to exclude periods consumed by litigation and other specified interruptions from the count, a concession to reality that in practice became a routine feature of case management, so that the published averages already carry the benefit of a clock that stops whenever the case is being fought about. Even with the clock stopped for an average of nearly four months per case, the measured time more than doubles the statutory outer limit, which is why the timeline evidence cannot be argued away as an artefact of counting method: the counting method is the generous one, and the cases overrun it regardless. The processes ending in orders for liquidation in the same data took an average of 499 days merely to reach the order that starts liquidation. Later editions of the Board’s newsletter series carry the cumulative average for resolution cases higher still: 621 days for the period ending March 2026 and 633 days for the period ending June 2026. A rising cumulative average is arithmetically unforgiving evidence, because each new period’s cases must be slower than the accumulated stock to pull the average up. The direction of travel is therefore not an artefact of early chaos being averaged in. The process has been getting slower as it has matured, at the same time as the stock of inherited legacy cases, the standard excuse for early slowness, has been worked through.

The delay begins before the statutory clock starts, in the gap between an application being filed and the tribunal admitting it, a gap the Code prescribed at fourteen days. The Economic Survey recorded the admission experience of operational creditors, whose applications require the tribunal to satisfy itself that a debt exists and is undisputed: an average of 468 days to admission in the financial year 2020-21 and 650 days in 2021-22. During that interval the defaulting management remains in control of the company, because the moratorium and the displacement of the board both date from admission, and the opportunities for value to leak out of the estate, through asset transfers, preferential payments and simple neglect, are at their maximum. Admission delay is thus not a preliminary administrative statistic. It is the period in which the Code’s central protection is absent and its central harm accumulates, and a system that takes on average well over a year to open the door of a process designed to run for six months has mislocated its bottleneck. Part of the cause is capacity in the tribunal system itself, its benches carrying the whole of company law alongside insolvency, which is examined in the profile of the National Company Law Tribunal and its appellate tribunal; part is the litigation load generated by the parties, appeals against admission, challenges to the committee’s decisions, and contests between classes of creditors, each individually arguable and collectively fatal to the calendar.

It is worth stating precisely what the timeline failure does and does not damage, because its effects are uneven. It does not falsify the realisation ratios, which are struck at approval whenever approval occurs, though it quietly erodes them, because the fair and liquidation values in those ratios are assessed at admission and the business being sold is worth less after two further years of suspended animation. It does damage the Code’s credibility as a threat, since a debtor who can spin a case out for years discounts the loss of control that is supposed to discipline him, and it damages small creditors and employees, for whom a claim frozen for six hundred days is a different injury from the same claim paid late by a solvent debtor. And it wholly defeats the design premise the Bankruptcy Law Reforms Committee stated most plainly, that speed is of the essence in insolvency because value is perishable. On the first of the article’s three honest measures the record is good. On the second it is poor, worsening, and documented in the regulator’s own publications, and no reading of the recovery ratios should be allowed to launder that finding away.

Individual cases give the averages a human scale, and the homebuyer litigation supplies the standard example. Jaypee Infratech, a real estate developer with tens of thousands of homebuyers awaiting flats, was admitted to the process in August 2017 on a bank’s application, in the Code’s first year. Its resolution was not finally approved until March 2023, when the tribunal approved the plan of Suraksha Realty, after a journey through competing plans, a Supreme Court-directed round of revised bidding, the intervention of the homebuyers as a class of financial creditors whose status Parliament had to legislate mid-case, and repeated appellate contests. Five and a half years for a single case is not an outlier in kind, only in visibility: the averages in the Board’s tables are built from thousands of less famous cases in which companies, creditors and employees waited comparable periods with no newspaper recording the wait. Nor is the delay evenly distributed in its causes. Admission contests consume the front of the case, the expression of interest and bidding consume the middle, and challenges to the approved plan, by dissenting creditors, by promoters, by tax authorities and by rival bidders, consume the end, so that a case can be prompt by the standards of each stage and still consume half a decade across the whole. That is the arithmetic by which a system of individually defensible hearings produces a collectively indefensible timeline, and it is why reforms aimed at a single stage, a faster admission or a stricter plan deadline, have repeatedly failed to move the total.

The stock of live cases tells the same story without any averaging. In the Board’s tables, the processes still ongoing at each reporting date number in the high hundreds and beyond, 1,865 remained open in the newsletter position for June 2026, and the Board’s age analysis of the live stock has persistently shown a majority of ongoing resolution processes already older than the statutory outer limit, with a substantial tail measured in years rather than days. A live stock skewed that way guarantees the future averages: the cases that will conclude in the coming periods are drawn disproportionately from the old stock, since the young cases have not had time to conclude, so the cumulative average completion time will keep rising even if every new case is handled faster than its predecessors. The appellate layer compounds the arithmetic and deserves separate notice, because the Indian system gives every disappointed participant a further forum by right. Admission orders, plan approvals, distribution decisions and liquidation orders are all appealable to the National Company Law Appellate Tribunal and then to the Supreme Court on questions of law, and the record shows each class of participant using the route: promoters against admission, dissenting creditors against distribution, unsuccessful bidders against approval, resolution applicants against the modification of their own plans. The early appellate traffic performed a systemic service, since it produced the settled law on admission, moratorium and committee power that later cases could rely on, and there is a real sense in which the first phase’s litigation was the price of the framework’s clarity. The later traffic is harder to defend in the same terms, because the governing principles were by then decided and the marginal appeal was usually about a particular case’s money rather than the law’s meaning, yet its timeline cost fell on the whole process. A striking feature of the averages is that they lengthened across exactly the period in which the law became most settled, which is the cleanest evidence in the record that the binding constraint is not legal uncertainty: the system knows what the rules are, and takes six hundred days to apply them anyway.

This is the statistical mechanism behind the rising series from 582 to 633 days, and it cuts both ways for interpretation. It means the deterioration is partly arithmetic, the working through of a backlog accumulated when tribunals, professions and procedures were new. It also means the backlog is real, is measured in live companies whose creditors and employees are waiting inside it, and will dominate the published record for years after the last reform is enacted, so that the timeline verdict of any assessment written within the system’s first decade is partly a verdict on a stock nobody can retrospectively accelerate.

The liquidation majority, and the composition answer

The count of outcomes is the statistic most often deployed against the Code, and it is real. In the ministry’s statement through December 2024, of 6,192 processes closed, 2,707 had been referred for liquidation against 1,119 that yielded resolution plans, with the balance of 1,236 closed on appeal, review or settlement and 1,130 withdrawn under Section 12A. The Reserve Bank’s Bulletin analysis of the Board’s data, covering 5,057 closed cases, distributed them in nearly the same proportions: about 16 per cent resolution plans, about 19 per cent withdrawn under Section 12A, about 21 per cent closed on appeal or review, and about 44 per cent ending in liquidation orders. Liquidation orders outnumber resolution plans by well over two to one in the cumulative stock. Anyone who tells the Code’s story without that fact is selling something, and anyone who stops at that fact has not asked what the liquidated population consisted of.

The composition evidence answers that question, and it is the Board itself that publishes it. The Reserve Bank’s analysis recorded that 77 per cent of the cases ending in liquidation were inherited from the earlier sick industrial companies regime or were already defunct units, in which substantial value erosion had taken place before admission. The Board’s figures for the position through September 2022 put the asset position of the two populations side by side: companies designated for liquidation had assets valued in aggregate at Rs 0.60 lakh crore on admission, against Rs 1.37 lakh crore for the companies that were resolved, and the average asset value of the liquidation-bound debtors was less than 8 per cent of the debt outstanding against them. These were, in the main, the dead of the previous regime, the companies that the Board for Industrial and Financial Reconstruction had held in suspended animation for years and in many cases decades, transferred into a Code whose repeal of the sick companies legislation, examined in the account of the SICA and BIFR repeal, made it the executor of that inheritance. For such companies a liquidation order is not the process failing to rescue a business. It is the first orderly exit the legal system ever offered them, and counting it as a failure of resolution is like counting a cremation as a failure of medicine.

The trend within the stock supports the composition reading for the early years and then begins to withdraw its support, and honesty requires both halves. The Reserve Bank’s analysis noted that resolutions as a proportion of liquidation orders rose from 21 per cent in the financial year 2017-18 to 45 per cent in 2022-23, a steady tilting toward resolution as the inherited stock cleared and the incoming caseload became companies that had failed under the Code’s own watch rather than under its predecessor’s. That improvement is the strongest evidence that the liquidation majority was a legacy phenomenon rather than a design verdict. But a defence built on composition has a shelf life. Each year the legacy share of admissions shrinks, so each year the outcomes describe the Code’s treatment of its own generation of failures more and the old regime’s less, and the timeline evidence over the same years moves the wrong way. The mature assessment is therefore double: the liquidation count, properly decomposed, does not convict the Code, and the excuse that acquits it is expiring, so that assessments made on later data will have to stand on the process’s merits for companies that were alive when they entered it.

Inside liquidation, the record has a further chapter that the order counts do not capture, which is the economics of the liquidation process itself. A liquidation order begins a second process, of asset identification, valuation, auction and distribution, which runs on its own clock: the Board’s figures for liquidation processes closed by final report put their average duration at 668 days across 1,529 closed processes, so the typical liquidated company’s creditors wait the better part of two further years after the order before the file closes. Within that process the regulations have been adjusted on the evidence, as the feedback section of this article records in the corporate context: liquidators were empowered to sell the corporate debtor as a going concern, and later to sell it together with its liabilities in defined circumstances, and a compromise or arrangement route under the Companies Act was opened for use inside liquidation, each reform an attempt to recover for creditors some of the going-concern value that the failure of resolution had supposedly extinguished. The going-concern numbers already quoted, 79 sales realising about Rs 3,674 crore against Rs 1.4 lakh crore of claims, measure how little those reforms have had to work with in the general run of cases. They are best read as evidence about the age and condition of the companies reaching liquidation rather than about the sincerity of the attempt, and they complete the liquidation picture: a majority outcome, dominated by companies that arrived dead, processed through a second lengthy procedure whose recoveries are small because there was little left to recover.

Two caseloads inside one statute: who files decides how cases end

The aggregates also hide a systematic difference between the cases brought by financial creditors and those brought by operational creditors, and the difference is itself implementation evidence, because it shows the statute being used for two different purposes by its two classes of user. Financial creditor applications, brought by banks and financial institutions against the large defaults, supply most of the value in the system and most of the cases that run their full course to a plan or a liquidation order. Operational creditor applications, brought by suppliers, service providers and other trade creditors, are far more numerous relative to their size, are concentrated in smaller defaults, and end disproportionately in settlement. CARE Ratings’ analysis of the Board’s data through September 2021 recorded that about 53 per cent of the cases withdrawn under Section 12A involved defaults of less than one crore of rupees, and that the leading recorded reason for withdrawal, in about 41 per cent of cases, was full settlement with the applicant. The Board’s later figures preserve the shape: of the 1,191 withdrawals under Section 12A counted by June 2025, 825 had been initiated by operational creditors. For a large class of trade creditors, in other words, the Code functions in practice as the most effective demand notice in Indian commercial law, a filing whose purpose is achieved when the debtor pays to make it go away, and which was never intended by either party to reach a resolution plan.

That dual use is neither wholly benign nor wholly abusive, and the record supports both readings at once, which is why it deserves separate statement. On the benign reading, the operational creditor’s filing is the settlement effect operating at small scale, and it disciplines thousands of commercial relationships in which late payment had been a business model; the supplier who is paid on the courthouse steps has been served by the statute exactly as a creditor-protection law should serve him. On the harder reading, pressed by the courts in a line of observations that the Code is a resolution mechanism and not a recovery tool, a process designed for the collective resolution of an insolvent firm is being consumed by bilateral debt collection, its tribunal capacity spent on disputes that were never about viability, and debtors are exposed to the catastrophic consequence of admission, loss of control of a going business, as leverage in arguments about contested invoices. The admission data cannot by itself decide between the readings, because a settled case looks identical in the tables whether the debt was genuine and desperate or tactical and disputed. What the data does establish is the scale of the second use, large enough that any reform of admission, of thresholds, or of the operational creditor’s route will change the system’s statistics dramatically without changing the resolution process at all, a further reason why the aggregates must be decomposed before they are debated.

The settlement effect: the largest consequence has no case file

Every statistic considered so far counts something that happened inside a process. The Code’s largest measured consequence happened outside any process, in the shadow of admission, and it is the effect the design most deliberately sought: a debtor who knows that a single admitted application ends his control of the company, permanently and within days, has a reason to pay that no earlier Indian regime ever gave him. The older systems let the defaulting promoter retain possession while forums deliberated, which made delay the debtor’s friend. The Code inverted the incentive, and the inversion is measurable in the applications that never became cases.

The Economic Survey of 2024-25 recorded that, from the Code’s inception through March 2024, 28,818 applications for initiation of the process, involving underlying defaults of Rs 10.2 lakh crore, were withdrawn before admission. The Board’s executive director, reporting the later position, put the number of cases settled prior to admission at 30,310 through December 2024, covering underlying defaults of Rs 13.78 lakh crore. These two figures are a dated pair, not a sum, and they belong to a different category from everything else in this article: no tribunal acted, no professional was appointed, no plan was approved, and the debt was addressed because the application existed. M S Sahoo, the Board’s first chairperson and the public face of the design in its early years, built the regulator’s account of the Code around exactly this behavioural change, arguing that the fear of losing the company had converted default from a negotiating position into an emergency. Critics are entitled to note that the claim comes from the institution with the strongest interest in making it, and that a withdrawn application records a settlement claimed by the parties rather than an audited payment. Both cautions are fair. They do not dissolve the phenomenon, because the withdrawal counts are compiled from tribunal filings, the quantum of default involved exceeds by a multiple everything realised inside the process, and no alternative explanation has been offered for why debtors began settling at the courthouse door in numbers the older forums never produced.

Post-admission settlement is a separate category and must not be merged into the same applause. Section 12A, inserted in 2018, permits withdrawal of an admitted application with the approval of ninety per cent of the committee of creditors, and through December 2024 the ministry counted 1,130 processes closed that way; a further 1,236 closed on appeal, review or settlement. The Board folds these outcomes, with the resolution plans, into its count of companies rescued: 3,485 of the 6,192 closures through December 2024, or 56 per cent, ended with the company surviving as a going concern rather than entering liquidation. Whether every settlement deserves the word rescued is a genuine question, since a withdrawal records that the applicant was satisfied, not that every creditor was paid or that the company was viable.

Does a settlement after admission count as an insolvency success?

Partly. A withdrawal under Section 12A, approved by ninety per cent of the committee of creditors by value, records that the creditors with money at stake preferred the settlement on offer to the process, the Code working as a bargaining framework. It does not record that every creditor was paid in full, or that the company was restored to health.

The settlement exit has itself required judicial policing, and the leading episode belongs in the record because it shows the category’s boundary being drawn. In the Byju’s litigation, an operational creditor’s application against Think and Learn Private Limited was admitted, a settlement was then reached with the promoter, and the appellate tribunal permitted the case to be withdrawn in exercise of its inherent powers, bypassing the Section 12A procedure and its requirement of ninety per cent committee approval. In GLAS Trust Company LLC v Byju Raveendran, decided in October 2024, the Supreme Court set that course aside, holding that once a process has been admitted and the committee of creditors constituted, withdrawal must follow the statutory route, because the proceeding has become collective and no pair of parties may dispose of it between themselves. The episode matters beyond its parties. A settlement culture in which the applicant and the promoter can close a case over the heads of the wider creditor body would convert the Code’s threat into a private instrument, valuable to whoever holds the filing and worthless as collective discipline, and the volume of Section 12A traffic documented in this article is exactly why the Court’s insistence on the collective route is an implementation fact and not merely a procedural one.

The balanced position is that the settlement effect, before and after admission, is the Code’s clearest systemic achievement and its least auditable one. It is clear because the volumes are enormous and the mechanism is the one the designers specified in advance, a prediction fulfilled rather than a rationalisation constructed afterwards. It is unauditable because a settlement’s terms are private, its durability unknown, and some share of the companies rescued by withdrawal will default again and return. An assessment that counts every settled rupee as a triumph overstates the record. An assessment that ignores the settlement effect because it appears in no recovery ratio understates it by more, and deletes the very consequence the statute was engineered to produce.

Where the record is genuinely weak

Three weaknesses survive every charitable reading, and the article’s verdict depends on not flinching from them. The first is the timeline failure already documented, which worsens in the regulator’s own series and which corrodes both of the Code’s other achievements, since delay discounts the threat and devalues the estate simultaneously. The second is the position of those at the wrong end of the distribution order. Section 53’s waterfall pays insolvency costs, then workmen’s dues and secured creditors who relinquish security, then a year’s wages and unpaid dues of employees other than workmen, and only later unsecured financial creditors, government dues and, towards the end, operational creditors’ remaining claims. In resolution the committee’s distribution has repeatedly left operational creditors, the suppliers and service providers whose unpaid bills triggered many cases, with a small fraction of what financial creditors received, and the complaint that the Code is a bankers’ statute in its incidence is not a misunderstanding of the design but a description of it. The numbers in completed liquidations show the floor of the problem: ICRA reported that in the 960 liquidations completed through March 2024, creditors realised about 4 per cent of their total admitted claims. The Board’s figures for liquidation processes closed by final report, 1,529 of them in the series reported in 2025, averaged 668 days to closure, so the smallest recoveries in the system are also among the slowest.

The third weakness is subtler and concerns what the resolution market rewards. A plan’s value depends on bidders, and the bidder pool is thinned by Section 29A’s exclusion of defaulting promoters and their connected persons, a disqualification whose integrity rationale is strong and whose price effect is real: fewer eligible bidders, in cases where the promoter was often the only party who valued the business as more than scrap, means lower realisations and some companies liquidated for want of an admissible buyer. The published record cannot apportion the haircut between value destroyed before admission, value destroyed by delay, and value never bid because the natural bidder was barred, and any commentator who assigns the whole gap to one of those causes is inventing a decomposition the data does not contain. What the data does contain, in the going-concern sales attempted inside liquidation, is a hint of how little residual value survives the full journey: the Economic Survey recorded 79 corporate debtors sold as going concerns in liquidation, with claims of Rs 1.4 lakh crore against realisations of about Rs 3,674 crore. By the time a company has passed through a failed resolution and a liquidation sale, the enterprise value the Code exists to protect is, in the typical case, already gone.

The distributional record inside liquidation deserves a sentence of its own, because the waterfall that the value ratios summarise has a human order. In a liquidation, the process costs and the secured creditors who relinquish their security come first, workmen’s dues for the twenty-four months before the liquidation date rank alongside them, and every other claim, the employee’s longer arrears, the small supplier, the unsecured lender, the government’s dues, queues behind for what the asset sales leave. In the asset-poor liquidations that dominate the caseload, what they leave is close to nothing, which is why the lived experience of the Code among employees and trade creditors of failed companies is so much bleaker than the value record’s averages: the averages are built on resolution cases, where going-concern value exists to share, while the typical individual’s encounter with the statute is a liquidation in which the waterfall runs dry several ranks above them. The Ahmedabad study’s finding that resolved firms raised their employee expenses by half is the other side of the same fact: survival is where the protection lies, and the statute protects people in proportion to whether their company is saved, a lottery whose odds the composition evidence has already given.

A fourth weakness belongs to scope rather than to performance within the process, and it belongs in an implementation record because it describes how much of the enacted statute the state chose to operate. The Code as passed covered individuals and partnership firms as well as companies, but the individual insolvency provisions were brought into force only for one class, personal guarantors to corporate debtors, by a notification that took effect in December 2019 and whose validity the Supreme Court upheld in Lalit Kumar Jain v Union of India in 2021. The guarantor limb has since generated a substantial caseload of its own, much of it shadowing the corporate cases, while the general framework for individual insolvency and bankruptcy, including the Fresh Start Process designed for debtors of minimal means, has never been notified at all. The commencement map itself is Article 55’s subject; the implementation point is narrower. Judged as a corporate statute the Code has a record, mixed but legible. Judged as the unified insolvency law its title proclaims, most of its personal insolvency territory remains unenacted in practice, and every aggregate in this article is a statistic about companies alone, a limitation that statements about “the Code’s record” routinely forget to carry.

A fifth weakness follows the successes out of the tribunal door. An approved plan is counted in the record at approval, but the record’s own later tables contain the companies whose approved plans were not implemented, where the successful applicant failed to infuse the promised funds or perform the promised restructuring and the company passed into liquidation after all, its resolution reversed in substance while standing in the statistics. The Board tracks these failures, and although they are a minority of approved plans, each one is doubly expensive: the creditors bore the full timeline of a resolution and then the full timeline of a liquidation, and the value that the plan process had preserved was spent in the interval. Allied to this is the concentration of the acquirer side of the market. A substantial share of approved plans has come from asset reconstruction companies, financial investors and a limited circle of strategic buyers rather than from a deep bench of industrial acquirers, which connects the plan-failure problem to the bidder-pool problem of Section 29A: the market for corporate control that the Code presupposes is thinner in practice than in design, and thin markets produce both lower prices at approval and weaker performance after it. Neither weakness appears in the four ratios, which is precisely why an implementation record that stopped at the ratios would be incomplete: a statute can score well on every measure taken at the tribunal door and still be failing in the years after the door closes behind its successes.

The last gap in the record is territorial. Modern corporate failure ignores borders, and the Code’s own provisions for it, sections empowering the government to apply the law to assets abroad by agreement and to issue letters of request to foreign courts, were left unoperationalised: no framework of reciprocal arrangements was built, and the UNCITRAL Model Law on Cross-Border Insolvency, whose adoption the Insolvency Law Committee recommended in 2018 with a draft chapter prepared for the purpose, was not enacted. The system therefore met its first great cross-border failure without a rulebook. When Jet Airways was admitted in 2019, insolvency proceedings concerning the same airline were running in the Netherlands, and the two systems had no statutory channel between them; the appellate tribunal’s answer, in the autumn of 2019, was to authorise an improvised cooperation protocol between the Indian resolution professional and the Dutch administrator, the first arrangement of its kind in Indian practice, under which the Indian process was recognised as the main proceeding and information and asset questions were coordinated by agreement. The improvisation worked well enough to become a small landmark, and it remains an improvisation: no later case has been given a statutory framework to inherit, and every cross-border failure since has faced the same blank page. For an implementation record the point is precise. The Code’s domestic machinery, whatever its delays, exists and functions; its international machinery exists only on paper, and the gap has been carried by judicial ingenuity rather than closed by the amendment process that repaired so many of the statute’s other defects.

The committee of creditors in practice: commercial wisdom and its discontents

The institution at the centre of the process generated its own implementation record, distinct from the statute’s. The committee of creditors was designed as a body of financial creditors voting by value, deciding the debtor’s fate by a sixty-six per cent supermajority for major decisions, and immunised by the courts against merits review of its commercial choices. In practice the committees of the large cases were dominated by public sector banks, whose officers carried a second set of incentives into the room: the personal exposure of a bank official who accepts a deep haircut to later vigilance and audit inquiry, a documented feature of Indian public banking that the committee’s designers, drawn from a tradition of assuming commercial actors behave commercially, did not price into the mechanism. The evidence of the resulting behaviour runs through the record without appearing in any ratio: plans renegotiated after approval votes, bidders pressed to improve offers across repeated rounds in ways that lengthened cases by months, and committees that preferred the visible safety of a liquidation they could blame on the process to a resolution whose haircut they would have to sign. The Board itself moved into this gap, circulating a discussion paper on a code of conduct for committees of creditors, and the Parliamentary Standing Committee’s recommendation of such a code, noted in the verdict section of this article, records the official system’s own conclusion that its central institution’s conduct was a variable in the outcomes, not a constant.

None of this displaces the countervailing evidence, which is that the same committees approved, across the full record, settlements and plans returning the liquidation value multiples this article has documented, and that the alternative decision-maker, the tribunal second-guessing commercial terms, was the arrangement whose failures under the older regime the Code was enacted to escape. The honest finding is narrower: the committee system works when its members’ incentives align with value maximisation and degrades where public accountability asymmetries make delay and liquidation personally safer than settlement, and the degradation shows up in the data as timeline, the variable over which the committee, unlike the tribunal, is rarely blamed. An assessment that assigns the whole timeline failure to bench capacity has not reckoned with the share of the calendar consumed inside committee rooms, in voting rounds, re-bidding demands and inter-creditor disputes between secured lenders of different ranks, a share no official series isolates and every practitioner recognises.

The feedback loop: what the record changed in the law

An implementation record is also a history of corrections, and the Code’s amendment history is unusually legible as a series of responses to its own early data. The distribution grievance came first. In the earliest large resolutions, committees of financial creditors exercised their discretion over plan proceeds in ways that left operational creditors with nominal sums, and the litigation that followed, culminating in the Supreme Court’s confirmation in the Essar Steel case that the committee’s distribution choices were its own, left the grievance answered in law and alive in politics. Parliament’s response, in the 2019 amendment, was not to disturb the discretion but to floor it: Section 30(2)(b) was recast so that operational creditors and dissenting financial creditors must receive at least what the waterfall would give them in liquidation, converting the liquidation value from an internal benchmark into a statutory minimum entitlement for the classes the early record had squeezed. The same amendment inserted the 330-day outer limit whose reading down in Essar Steel this article has recorded, a pairing that captures the period exactly: the legislature reacting simultaneously to distributions it disliked and to a timetable it could see slipping, and the court then moderating the timetable reaction in the very judgment that confirmed the distribution.

Later corrections followed the same pattern of evidence first, legislation after. The homebuyer’s status, litigated through the courts while thousands of allottees waited outside the creditor classes the original text had given them, was regularised by amendment once the caseload had made the gap undeniable. The pandemic produced the threshold increase and Section 10A, emergency corrections whose statistical consequences this article has flagged. The pre-packaged process for micro, small and medium enterprises, introduced in 2021, was a design correction aimed at a documented misfit, the unsuitability of the full process for small firms, though its uptake in its early years was minimal, with the Board’s published counts of admissions under the framework standing in single figures across its first two years, a reminder that supplying a better-fitting procedure does not by itself create the professional capacity or creditor confidence the procedure needs. The pre-pack design asks financial creditors to negotiate a restructuring while the existing management stays in possession, the inverse of the main process’s founding suspicion of defaulting promoters, and creditor institutions formed by a decade of promoter misconduct were slow to extend that trust; the episode measures the distance between a legislature’s willingness to correct a design and a market’s willingness to use the correction. The pattern across all of these is worth naming, because it distinguishes this statute’s implementation from the stagnant norm the Code displaced: the corrections came, they came in response to published evidence and decided cases, and several of them worked. A framework that amends itself on evidence is not thereby vindicated in its outcomes, the timeline record stands against it, but the feedback loop is a real institutional property, visible in the record, and any assessment that freezes the Code at its 2016 text is assessing a statute that no longer exists.

The sectoral record: an aggregate built from unlike industries

The Code’s caseload is not a sample of the Indian economy; it is a sample of the Indian economy’s failures, weighted by where credit had flowed and where the preceding decade’s distress had concentrated. Manufacturing supplies the largest share of admissions, reflecting the industrial bad-loan cycle the Code was designed in the shadow of, and real estate and construction supply the most politically visible share, counted by the Economic Survey at more than 1,500 admissions and about 21 per cent of the total through March 2024. The sectors behave differently inside the process, and the aggregate ratios are averages over behaviours that have little in common. A manufacturing company carries plant, inventory and an order book, assets a resolution applicant can value and a going concern a buyer can operate; its case turns on industrial viability. A real estate company carries land parcels, approval risk and thousands of retail homebuyers whose claims are flats rather than money; its case turns on completing construction, and its resolution arithmetic, in which the customers are simultaneously the creditors, has no counterpart elsewhere in the tables. Of 891 real estate companies admitted through March 2024 whose processes had yielded outcomes, the Survey recorded 133 resolutions, 47 withdrawals and 319 liquidation orders, a resolution share far below the system’s, produced by an industry in which the asset is often a half-built project whose completion value exists only if the resolution itself supplies fresh money.

Services and trading companies sit at the other end of the asset spectrum, and their cases dominate the liquidation statistics’ hollow core: businesses whose value was contracts, relationships and staff, all of which dissipate on admission, leaving the process to distribute nearly nothing through a full statutory ceremony. The implementation lesson the sectoral split teaches is that a uniform process applied across unlike asset structures will produce a misleading aggregate and sector-specific pathologies simultaneously. The system’s designers anticipated part of this, in the project-wise and homebuyer amendments that real estate forced upon the statute, and the pre-pack framework for micro, small and medium enterprises was a later admission that the standard process fits small manufacturing and service firms badly. The sectoral record is therefore the strongest internal evidence for differentiated insolvency tracks, and the weakest part of the case for judging the Code by any single ratio: the same number that records a steel plant saved records a flat project abandoned, and the average of the two is a fact about the mixture, not about the institution.

Real estate also supplies the clearest example of a caseload being deliberately re-engineered mid-stream. Once homebuyers had been given financial creditor status, individual allottees could and did initiate processes over single delayed projects, and the filing pattern alarmed the government into a threshold: an amendment effective in 2020 required a real estate application by allottees to be filed jointly by at least one hundred allottees or ten per cent of the total number in the project, whichever was less, a gate upheld by the Supreme Court in Manish Kumar v Union of India. Whatever the merits of protecting developers from single-buyer filings, the episode belongs in the implementation record as a change in the population itself: admissions data before and after the threshold describe differently filtered caseloads, and the same is true of the pandemic threshold increase for all sectors. Three times within the Code’s first five years, in the homebuyer amendment, the threshold notifications and the pandemic suspension, the state altered who may enter the process and on what terms. The ratios survived those alterations with their meanings intact only for readers who track the gates alongside the outcomes, which is the methodological burden this whole article has been carrying, and the sectoral evidence is where the burden bites hardest.

The supply side: the profession and machinery the record presupposes

Implementation evidence includes the parts of the machine that produce no headline ratio: the people and institutions through which every case must pass. The Code created a regulated profession where none had existed, the insolvency professionals who take control of the debtor on admission, run the process, invite and evaluate plans and answer to the Board for their conduct, and several thousand professionals have been registered, examined and subjected to a disciplinary regime within the statute’s first years. It created registered valuers, whose paired estimates of fair value and liquidation value generate two of the four ratios this article is built on. It provided for information utilities, the repositories whose records of default were meant to make admission a near-ministerial act, and in practice a single utility, National e-Governance Services Limited, has carried that function, with the consequence that the evidentiary shortcut the designers intended operates only where a debt happens to be recorded with it. Each of these is an institution that had to be conjured into existence by regulation, examination and market formation, and the record of the early phases, in which tribunals improvised around missing machinery, cannot be understood without crediting that the machinery mostly arrived.

The cost evidence belongs in the same ledger, because a process can be effective and still price out the small cases it was meant to serve. The Board’s figures put the average cost of a resolution process at 1.11 per cent of liquidation value and 0.63 per cent of the value realised under plans, which are low ratios by international comparison and were cited by the Board’s leadership in the regulator’s claim that India had built one of the lowest-cost insolvency jurisdictions. The averages conceal the distribution that matters: costs contain a large fixed element of professional fees, tribunal process and valuation, so the burden as a share of value is far heavier for a small company than for a giant, and the raising of the initiation threshold to one crore of rupees in 2020 quietly withdrew the smallest debtors and creditors from the system altogether, improving the average economics of the caseload by narrowing who may use it. That trade, between the purity of the aggregate record and the breadth of access to the statute, is one the headline statistics never disclose, and it is a recurring shape in this field: the record improves partly because the process improved, and partly because the population it serves was redefined to exclude the cases in which it performed worst.

The comparative check: the routes the Code displaced

A record is only legible against the counterfactual, and for the Code the counterfactual is the system it replaced: a creditors’ remedies landscape of suits, debt recovery tribunals, and enforcement under the SARFAESI Act by secured creditors, running alongside a sick companies regime that suspended recovery while offering no resolution. The World Bank’s Doing Business assessments measured the old world’s pace: resolving insolvency in India took 4.3 years in the assessment published before the Code, while the 2020 edition, assessing the reformed framework, measured 1.6 years for its model case. Those figures describe a standardised hypothetical rather than the caseload, and the newsletter averages in this article, running well above even the older Doing Business measure for the model case, warn against treating either as the lived timeline. What the comparison establishes is the direction of the design change, not its delivered magnitude: the Code made the formal process radically faster on paper, and the gap between the paper and the averages is the implementation failure, measured against the Code’s own promise rather than against its predecessor’s despair.

The same assessments carried a recovery comparison, and it illustrates why this article insists on bases. The pre-Code Doing Business assessment put recovery for creditors at 25.6 cents on the dollar for its standardised case in an economy where enforcement ran through the courts and the debt recovery tribunals over years; the 2020 edition put the figure at 71.6 cents for the standardised case under the new framework. Set beside the Board’s real-world figure of roughly a third of admitted claims, the 71.6 cents looks like a fantasy, and it has been quoted triumphantly by the Code’s defenders and derisively by its critics. Both reactions miss what the number is: the return in a single hypothetical case, specified by the assessors, in the country’s largest business city, assuming a going-concern sale of a particular hotel business. It measures the law on the books for a well-formed case, and on that question it is genuinely informative, because the old books scored 25.6. It does not measure the caseload, with its defunct companies, its admission queues and its five-year contests, and using it as if it did is the comparative version of the denominator error, a model case mistaken for a population. The correct use of the pair is the one this section makes: the design improved radically on paper, the delivered record improved materially and unevenly in fact, and the distance between the two sentences is the size of the implementation problem.

The flow of actual bank recoveries gives the comparative check its harder edge. The Reserve Bank’s Report on Trend and Progress of Banking, as placed before the public by the Ministry of Corporate Affairs, recorded that in the financial year 2023-24 the Code accounted for 48 per cent of all recoveries made by scheduled commercial banks, ahead of the SARFAESI route at 32 per cent, the debt recovery tribunals at 17 per cent and Lok Adalats at 3 per cent. Within a few years of commencement, in other words, the new process had become the banking system’s principal recovery channel by value, displacing the enforcement route examined in the guide to the SARFAESI Act 2002 from the position it had held for a decade and a half. That is a market verdict, delivered by the creditors free to choose their forum, and it corroborates the behavioural account: lenders file under the Code because the threat it carries settles cases, and they recover through it because the resolution market pays more than piecemeal enforcement against a carcass. The comparison has a limit that must be stated with it. Channel shares measure where recovered money flowed, not how much value the economy preserved, and a channel can dominate recoveries while its timelines decay, which is the position the timeline evidence describes.

The balance-sheet cross-check: what happened to bank bad loans

The widest claim made for the Code is not about its cases at all but about the banking system: that the era of evergreened industrial bad loans ended because the Code made default consequential. The balance-sheet facts to be explained are not in dispute. The Reserve Bank’s reporting records the gross non-performing asset ratio of scheduled commercial banks at 11.2 per cent at its March 2018 peak and at 2.3 per cent by March 2025, a transformation of reported asset quality within the Code’s operating life. The dispute is over attribution, and it is a genuine dispute, because at least four mechanisms were running at once: recoveries through the channels this article has tabulated, of which the Code is the largest; technical write-offs, which remove bad loans from the ratio by acknowledging the loss rather than recovering it, and which the finance ministry has stated in Parliament amounted to some Rs 16.35 lakh crore of loans written off across the decade to 2023-24; fresh slippages slowing as bank underwriting tightened and the asset quality review generation of loans worked through; and the simple growth of the loan book, which dilutes a ratio whose numerator is static. A claim that the Code caused the fall in the ratio must survive contact with the write-off figures, which alone exceed total insolvency realisations several times over, and it cannot.

What survives is the narrower and more defensible claim about behaviour and pricing rather than about the ratio itself. The Code changed what a default means for the borrower’s controllers, converting it from an occasion for restructuring negotiations into a credible risk of losing the company, and that change is the one the pre-admission settlement figures measure directly: tens of thousands of defaults settled before admission, on the Board’s later count 30,310 cases involving Rs 13.78 lakh crore of underlying default by December 2024, are loans on which the banking system recovered or restructured without a case, under a threat that did not exist in the same form before 2016. Bankers’ own accounts of the period, and the Reserve Bank’s analysis of the settlement pattern, treat that shadow as the Code’s largest banking effect, larger than its case recoveries and impossible to see in the NPA ratio, which records only the defaults that happened and not the ones averted or settled early. The cross-check therefore returns a split verdict consistent with everything else in this article: the Code’s measurable contribution to bank recovery is real and channel-leading, the claim that it cleaned the balance sheets is an attribution the data does not support, and the truth, a behavioural effect visible in settlements and in the changed conduct of large defaulters, sits between the slogan of transformation and the dismissal that nothing changed.

How to quote the insolvency evidence without being wrong

The practical yield of this record, for the journalists and researchers who quote it, compresses into a discipline of four questions asked of any insolvency number before it is used. The first is: what is the denominator? A realisation figure against admitted claims, against liquidation value and against fair value will differ by a factor of five for the same cases, and a percentage carried without its base is not a shortened fact but a false one. The second is: what is the period? The cumulative stock through any date mixes the legacy inheritance with the Code’s own caseload and the slow early tribunals with the slower later ones, while a single financial year’s flow is hostage to whichever large cases concluded within it; the two answer different questions and quoting one as the other manufactures trends in both directions. The third is: what event is being counted? Plan approvals are commitments, liquidation orders are beginnings, and closures by settlement or withdrawal are survivals of uncertain depth, and a count that treats the three as equivalents of recovery has abandoned measurement for mood. The fourth is: what happened outside the count? The settlement effect means the process’s influence is systematically larger than its throughput, so a purely internal audit of any insolvency system will underrate a regime whose main product is deterrence, just as a purely celebratory audit will overrate one whose settlements are counted as cures.

A worked example shows what the discipline does to a live claim. Take the statement that creditors suffered haircuts of nearly seventy per cent under the Code, a claim built honestly from the Economic Survey’s computation of a roughly 69 per cent haircut against admitted claims for the resolution cases to September 2024. The denominator question shows that the same cases returned 161 per cent of liquidation value, so the claim and its opposite, that creditors did far better than breaking up the assets, are both arithmetic drawn from one dataset. The period question shows that the claim describes a cumulative stock of concluded cases heavy with the legacy failures of the first phase, not the marginal case of any recent year. The stage question shows that the claim is about resolution cases only, a seventh or so of closures, and that the liquidation population’s far deeper losses sit outside it, so the claim understates total creditor losses while overstating resolution losses. And the counterfactual question asks what those admitted claims were worth on the day of admission, when they were claims on companies that had typically been in default for years, were carried on bank books at steep provisions, and faced a liquidation value equal to about a fifth of their face. A claim admitted at one hundred rupees against a company whose assets would fetch twenty in a sale is not a hundred rupees of value destroyed by a process that returns thirty-two; it is twelve rupees of value created over the alternative, recorded in the tables as a sixty-eight per cent haircut. The example is stylised, but every element in it is a ratio the official record publishes, which is the point: the honest and the misleading account of this statute are assembled from the same parts, and the assembly is the argument.

None of this discipline is a plea for agnosticism. The evidence, handled with its denominators and periods attached, supports firm conclusions, and an assessor who has done the work is entitled to reach them. It is a plea for recognising that in this field the denominator is the argument: the haircut debate has lasted years not because the parties dispute the arithmetic, which is common ground in the Board’s tables, but because each side selects the ratio that converts its prior view into a headline. The general account given in this article is legal information and analysis of a published statistical record, and it is not advice on any individual insolvency, claim or settlement, on which the position turns on the documents and the forum. A reader who applies the four questions to the next insolvency headline they meet, from whichever direction it arrives, will find that most dissolve on contact, and that the ones which survive are usually the ones the Board published with the base and the period attached.

The verdict on the record

The Code was given an objective, and the honest verdict is that it has met the objective unevenly in a pattern the evidence specifies precisely. On value, it has done what it promised: resolution has returned around one and two-thirds to one and seven-tenths times the liquidation value of the same assets across successive official cuts, and close to nine-tenths of assessed fair value, which means the committee of creditors, given control, has repeatedly extracted going-concern prices from estates the alternative process would have sold for scrap. On time, it has failed against its own text and the failure is compounding: a 330-day outer limit that courts would not let stand as mandatory, admission queues measured in hundreds of days, and a cumulative average completion time that has risen from 582 days in the September 2024 data to 633 days in the June 2026 newsletter, moving away from the statute every year. On behaviour, it has succeeded beyond what its throughput suggests: tens of thousands of debtors settling before admission against defaults counted in lakh of crore, and the banking system’s recovery flow reorganised around the threat the Code carries. The composition defence of the liquidation majority is sound for the years it describes and is expiring as the legacy stock clears, so the system’s defenders should spend less time citing it and its critics less time ignoring that it was ever true.

The institutional system has, to its credit, published the evidence of its own shortcomings and occasionally acted on it. The Parliamentary Standing Committee on Finance, in its 2021 report on the implementation of the Code, recorded the timeline slippage and the frequency of liquidation outcomes, pressed for the vacancies in the tribunal system to be filled, and recommended a professional code of conduct for committees of creditors, whose commercial decisions the courts had placed beyond review without any corresponding discipline of standards. Successive amendment exercises have taken up fragments of that agenda, and the recurring official response to the timeline evidence has been capacity: more benches, more members, dedicated insolvency strength. The record supports that diagnosis in part, since admission queues are a capacity phenomenon, but it also shows delay distributed across every stage and generated by parties as much as by benches, which is why the averages have continued to rise through periods of acknowledged reform effort. A verdict on implementation that stopped at “add capacity” would be repeating the system’s own preferred explanation rather than testing it, and the test the data applies, stage by stage, is the one the next round of reform will be judged against.

One quantity in the record deserves to be read as the hinge on which the rest turns, and it is the simplest of all: in every one of the cases that ended in a resolution plan, control of the company passed away from the people who had controlled it into default. That transfer is not a side effect of the process; it is the process, the event the entire architecture exists to make credible, and its repetition across more than a thousand companies, including some of the most prominent industrial houses in the country, is what converted the Code from a text into a fact of commercial life that boards and promoters price into their decisions. The pre-Code order’s defining feature was the separation of default from consequence for controllers, who kept their companies through restructuring after restructuring while creditors absorbed the losses in instalments. Measured against that baseline, the single most consequential implementation fact is that the separation ended: not for every defaulter, since liquidation and settlement have their own distributions of pain, and not at acceptable speed, since the timeline evidence stands; but as a general expectation in the minds of the class whose behaviour the statute aimed to change. Every other measure in this article, the ratios, the settlements in the shadow, the channel shares, is downstream of that expectation, and assessments of the Code that omit it are weighing the cargo and ignoring the engine.

What the record does not support is either of the slogans. It does not support failure, because a failed statute does not beat its own counterfactual on value in every published period, reorganise creditor behaviour at national scale, and become the principal recovery channel of the banking system within a few years. It does not support triumph, because a triumphant statute does not take twice its own maximum time and lengthen, strand operational creditors at the wrong end of its waterfall, and leave the companies in its completed liquidations recovering four paise in the rupee. The defensible formulation is narrower and more useful: the Insolvency and Bankruptcy Code is a value and behaviour reform that has worked, delivered through a timeline that has not, and its future assessment will turn on whether capacity and litigation discipline can be made to serve the design before the design’s credibility, which is the asset its settlement effect runs on, is spent by the delay its own statistics record.

For the reader returning to this record in later periods, four series will show first whether the pattern is changing, and they are the ones to watch in the Board’s successive newsletters. The first is the average completion time for resolution processes, whose rise from 582 to 621 and then 633 days across the published cumulative cuts is the single clearest deterioration in the record; a plateau in that series would be the earliest sign that capacity and discipline reforms are biting, and its continued rise would falsify them. The second is the admission interval, the gap between filing and admission against the fourteen days the statute prescribes, because admission delay is the stage least defensible in principle and the one whose improvement requires nothing but listing discipline and bench strength. The third is the composition of outcomes among companies that were not inherited from the older regime, the population on which the mature Code will be judged, where a rising resolution share with stable liquidation value multiples would establish that the process serves live companies as designed. The fourth is the pre-admission settlement count, the measure of the statute’s shadow: if the behavioural effect is durable, settlements should remain large even as tribunals speed up, and if the shadow is a novelty that fades as promoters learn the system’s actual tolerances, the settlement series will fade first and the verdict offered here would need revision in the direction of the critics. The record, kept on those four series with their denominators attached, will write the next assessment more reliably than any commentary, this one included.

Frequently Asked Questions

Q: How much do creditors actually recover under the insolvency process?

In cases that produced approved resolution plans, creditors have realised roughly a third of their admitted claims across successive official cuts of the record: the Indian Institute of Management Ahmedabad study computed 32 per cent of admitted claims, and the Insolvency and Bankruptcy Board’s statement through March 2025 recorded about 32 per cent on realisations of Rs 3.89 lakh crore. The same plans returned about 170 per cent of the liquidation value of the assets and between about 86 and 93 per cent of their fair value, depending on the period and the cases in which fair value was assessed. In completed liquidations, by contrast, ICRA reported realisation of only about 4 per cent of admitted claims through March 2024.

Q: Why are the haircuts under the insolvency process so large?

Because the haircut is measured against admitted claims, and admitted claims are the debts of companies that had already lost most of their value before any case began. A company’s borrowings accumulate over years of decline, while its assets are valued at the point of failure, so the gap between the two largely records pre-insolvency destruction rather than loss inside the process. The composition of the caseload deepens the effect: many admitted companies were defunct or legacy cases from the sick companies regime, and creditors in the defunct cases that were resolved realised under a fifth of their claims. Delay then erodes what value remains, because the business being sold is worth less after years of suspended animation.

Q: How long does a corporate insolvency resolution process really take?

Far longer than the statute allows. The Code prescribes 180 days, extendable by 90, with an outer limit of 330 days including litigation time, but the Economic Survey of 2024-25 reported the Board’s figure of an average 582 days for the 1,068 processes that had yielded resolution plans through September 2024, excluding a further 116 days on average that adjudicating authorities had excluded from the count. Later editions of the Board’s newsletter series record the cumulative average rising to 621 days for the period ending March 2026 and 633 days for the period ending June 2026. The time taken merely to get a case admitted comes on top of all of these figures.

Q: Do more cases end in liquidation than in resolution?

Yes, by count of orders. The Ministry of Corporate Affairs’ statement through December 2024 recorded 2,707 processes referred for liquidation against 1,119 that yielded approved resolution plans. Two qualifications define what that means. First, the remaining closures were survivals: 1,236 processes closed on appeal, review or settlement and 1,130 were withdrawn under Section 12A, so 3,485 of 6,192 closed cases, about 56 per cent, ended with the company rescued rather than liquidated. Second, a liquidation order is not a completed liquidation; it opens a separate process of sale and distribution that itself runs for years, so the two counts describe different stages of different journeys.

Q: Does the recovery rate measure the insolvency law’s success?

No. The Code was designed as a resolution law, not a recovery law: its objective, in the Bankruptcy Law Reforms Committee’s framing, was to keep viable firms operating under new control, close unviable ones quickly, and change debtor behaviour by making loss of control on default a credible threat. Recovery is a consequence of those things, not the target. The measures that test the design are whether resolution beats liquidation value for the same assets, whether the process completes fast enough to preserve value, and whether debtors settle before admission. On those three measures the record is respectively strong, poor and very strong, which no single recovery percentage can express.

Q: How much debt is settled before an insolvency case is even admitted?

A very large quantum, and it is the Code’s largest measured effect. The Economic Survey of 2024-25 recorded that from the Code’s inception through March 2024, 28,818 applications involving underlying defaults of Rs 10.2 lakh crore were withdrawn before admission. The Board’s executive director reported the later position as 30,310 cases settled prior to admission through December 2024, covering underlying defaults of Rs 13.78 lakh crore. These figures are kept separate in this article from settlements after admission, including withdrawals under Section 12A, because the mechanism differs: pre-admission settlement is a debtor paying to prevent the loss of control that admission would trigger automatically.

Q: Why does admission to insolvency itself take so long?

The Code prescribes fourteen days for the tribunal to admit or reject an application, but the Economic Survey recorded average admission times for operational creditor applications of 468 days in the financial year 2020-21 and 650 days in 2021-22. The causes are layered: the tribunal must satisfy itself that a debt exists and, for operational creditors, that it is not genuinely disputed, which invites the debtor to contest the evidence; promoters file objections and interlocutory applications that must each be heard; and the tribunal’s benches carry the whole of company law alongside insolvency with limited strength. The cost of the gap is high, because until admission the moratorium has not begun and the defaulting management remains in control of the assets.

Q: Has the insolvency law revived any companies?

Yes, in large numbers, including companies the previous regime had written off. Through December 2024, 1,119 processes had yielded approved resolution plans that kept the corporate debtor alive under new ownership, and the ministry recorded that around 39 per cent of those cases had previously been before the Board for Industrial and Financial Reconstruction or were defunct, which means several hundred formal revivals of companies already classified as dead. The Indian Institute of Management Ahmedabad study of resolved firms found their sales grew by 76 per cent in the years after resolution, employee expenses rose by 50 per cent, and the firms reached operational break-even by the third year.

Q: Why do different sources report different recovery percentages under the Code?

Because the percentage has three possible denominators and at least two possible periods, and sources rarely state theirs. Against admitted claims, realisation in resolved cases is roughly a third. Against liquidation value it is about 170 per cent, and against fair value somewhere between about 86 and 93 per cent, the range reflecting different periods and case subsets. A fiscal-year cut differs again from the cumulative stock: ICRA reported the average haircut in plans approved within the year 2023-24 at 73 per cent, much deeper than the cumulative position, because a single year’s conclusions are dominated by whichever large cases finished in it. Every figure can be correct and the comparisons still false.

Q: What is the difference between a liquidation order and a completed liquidation?

A liquidation order is the tribunal’s direction that the company be wound up, passed at the end of an unsuccessful resolution process or where the committee of creditors so decides; it is the start of liquidation, not its end. A completed liquidation is the conclusion of the liquidator’s work of selling assets, adjudicating claims and distributing proceeds, ending in a final report and dissolution. The two events are years apart: the Board’s figures reported 1,529 liquidation processes closed by final report at an average of 668 days. Commentary that counts liquidation orders as if they were completed liquidations overstates how much of the caseload has reached any final outcome at all.

Q: How many admitted cases are withdrawn under Section 12A or closed by settlement after admission?

Through December 2024, the Ministry of Corporate Affairs counted 1,130 processes withdrawn under Section 12A and a further 1,236 closed on appeal, review or settlement, out of 6,192 closed cases. Section 12A, inserted in 2018, allows an admitted application to be withdrawn with the approval of ninety per cent of the committee of creditors by voting share. Together with the 1,119 resolution plans, these settlement routes are why the Board counts 3,485 companies, about 56 per cent of closures, as rescued. They are distinct from pre-admission settlements, in which no case was ever admitted, and the two categories must never be added together as one settlement figure.

Q: Why do operational creditors recover so little compared with banks in insolvency cases?

The distribution order explains most of it. In liquidation, Section 53 pays insolvency costs first, then workmen’s dues and relinquishing secured creditors, then a year’s wages of other employees, and operational creditors’ claims rank far below unsecured financial creditors and government dues. In resolution, the committee of creditors is composed of financial creditors, and the distribution it approves has repeatedly given operational creditors, the suppliers and service providers whose unpaid invoices triggered many cases, a small fraction of what banks received, subject to the floor protections the courts have required for dissenting and operational creditors. The outcome is a feature of the statute’s hierarchy, criticised as such since enactment, not an accident of individual cases.

Q: Do companies that come through a resolution plan perform better afterwards?

The best available evidence says yes. The Indian Institute of Management Ahmedabad study of firms resolved under the Code, published through the Board in August 2023, tracked their post-resolution accounts and found sales growth of 76 per cent, a 50 per cent rise in employee expenses, growth in total assets and capital expenditure, and the resolved firms reaching operational break-even by the third year, with profitability aligning towards industry benchmarks. The finding supports the Code’s premise that transferring a viable business to new ownership restores it, rather than merely repricing its debts. It rests on one institutional study of the firms that survived resolution, so it describes the resolved population and cannot speak for companies that entered the process and failed.

Q: Which industries account for the largest share of insolvency admissions?

Manufacturing has consistently been the largest sector in the Board’s sectoral distribution of admissions, followed by real estate and construction. The Economic Survey of 2024-25 recorded that more than 1,500 real estate companies had been admitted through March 2024, about 21 per cent of all admissions, and that about one in four cases settled after admission also came from that sector; of the companies resolved by that date, 133 were real estate companies, about 15 per cent of resolutions. The sectoral pattern matters to the evidence because asset-heavy manufacturing cases tend to produce the strongest realisations against liquidation value, while real estate cases raise distinct problems of homebuyer claims that the aggregate ratios conceal.

Q: Can a company be sold as a going concern after a liquidation order?

Yes. The liquidation regulations permit the liquidator to sell the corporate debtor, or its business, as a going concern, preserving the enterprise, its employees and its licences as a unit rather than breaking the assets up piecemeal. The Economic Survey recorded 79 corporate debtors that continued operations through going-concern sales in liquidation, against total claims of about Rs 1.4 lakh crore and realisations of about Rs 3,674 crore. The mechanism matters because it shows the system’s preference for enterprise survival operating even after resolution has failed, but the figures also show its limits: by the time a company has passed through a failed resolution process into liquidation, the going-concern value remaining to be sold is usually a small residue of the original business.

Q: What does delay cost a company and its creditors while an insolvency case runs?

Delay consumes the asset the process exists to distribute. A distressed company’s value depends on contracts, customers, employees and licences that decay while its fate is undecided: suppliers tighten terms, buyers defect, skilled staff leave, and maintenance is deferred, so the fair and liquidation values assessed at admission overstate what a buyer will pay years later. The cost falls unevenly. Financial creditors discount the delay into the price they accept; employees and small operational creditors, whose claims are frozen for the duration, bear it as lost livelihood and working capital. Delay before admission carries a further cost, because the defaulting management remains in control and value can be transferred out of the estate before the moratorium begins.

Q: How much of the insolvency timeline is consumed by litigation?

A substantial and deliberately unquantified share, unquantified because the Board’s published averages do not decompose the total by cause. What is documented is the structure: the 2019 amendment wrote litigation time into the 330-day outer limit precisely because appeals against admission, challenges by promoters, and contests between creditor classes had become the standard means of consuming the clock, and the Supreme Court in the Essar Steel case then declined to make that limit mandatory, so litigation can extend a process beyond it in exceptional cases. The Board’s averages do record that adjudicating authorities excluded an average of 116 days per resolved case from the statutory count through September 2024, which is a partial, tribunal-certified footprint of the time the formal process itself disowned.

Q: Were most companies ordered into liquidation already defunct before the process began?

A large majority of them, on the regulator’s own analysis. The Reserve Bank’s Bulletin analysis of the Board’s data recorded that 77 per cent of the cases ending in liquidation were inherited from the Board for Industrial and Financial Reconstruction regime or were already defunct units whose value had eroded before admission. The Board’s figures through September 2022 showed the liquidation-bound companies held aggregate assets of Rs 0.60 lakh crore on admission against Rs 1.37 lakh crore for the companies that were resolved, with average asset value under 8 per cent of the debt outstanding. For that population a liquidation order was an orderly exit from a death that had already occurred, not a failure of rescue.

Q: How reliable are the published insolvency statistics themselves?

Reliable in pattern, approximate in point detail, and honest reading requires knowing why. The Board’s quarterly newsletters are compiled from reports filed by insolvency professionals across thousands of cases, and cases are sometimes reported late, so cumulative totals are revised between editions and a figure quoted from one quarter may shift slightly in the next. Different official publications also cut the record differently: a ministry statement, an Economic Survey chapter and a newsletter table may cover different end dates and case subsets, which is why the fair value realisation appears variously at about 86, 88 and 93 per cent in sources that are all official. The stable findings survive every cut; the decimal points do not, and should never be quoted without their source and period.

Q: How do recoveries under the Code compare with the older recovery channels banks used?

By the banking system’s own accounting, the Code became the dominant channel quickly. The Reserve Bank’s Report on Trend and Progress of Banking recorded that in the financial year 2023-24 the Code accounted for 48 per cent of all recoveries by scheduled commercial banks, ahead of enforcement under the SARFAESI Act at 32 per cent, the debt recovery tribunals at 17 per cent and Lok Adalats at 3 per cent. The comparison measures where recovered money flowed rather than rates of recovery within each channel, and it partly reflects the settlement effect, since filings under the Code induce payment even where no case is admitted. It nonetheless records a genuine reordering: the process creditors feared debtors would ignore became the route through which most bank recovery value moved.