On August 27, Vijay Mallya—a fugitive economic offender resident in Britain since 2016—trolled the system by posting a complaint on X. Industrialist Subhash Chandra had just settled dues of ₹22,006.57 crore for ₹6.5 crore under the Insolvency and Bankruptcy Code. “If true,” Mallya wrote, “many congratulations to my friend Subhash. Banks and government have admitted having recovered ₹14,100 crore from me against a judgement debt of ₹6,203 crore. Many more borrowers have settled at a fraction. Indian debt resolution justice, I presume. No media questions.” The irony is that a man synonymous with bad loans was complaining that too much had been recovered from him.
Every few months, a screenshot does the rounds: a list of ‘wilful defaulters’, a running total of loans ‘written off’, a caption asking why a small borrower loses his cow for a ₹20,000 default while a tycoon boards a private jet owing ₹9,000 crore. The question is never only about who fled. It is how much came back.
The triggers for outrage are mostly driven by headlines—the latest is a conflation of the Chandra case, which technically is from the guarantor universe, and the stated claim that no bank has a net non-performing asset ratio above 1 percent of the total loans. Gross NPAs—total defaults before provisions to cushion the bank—which peaked at 14.58 percent of advances in March 2018, fell to 1.93 percent by March 2026. The numbers suggest that the crisis of a decade ago has been closed out.
The ratio reflects the shifting onus of recovery. Mark the sequence. A loan turns non-performing; the bank provisions against it fully within four years, as RBI rules require, before the board writes it off. The finance ministry clarifies it is “an accounting procedure” and that the borrower remains liable for repayment. Whatever recovery happens—through a tribunal, a settlement, an asset sale—happens sometimes after years, sometimes never. The net NPA figure and the recovered rupee are two different clocks, and only one of them has been running fast.
The write-off saga shows up at scale in Parliament. Around ₹19 lakh crore of loans have been written off over the past 11 years. The lagging recoveries are barely a third of the write-offs. In the past 5 years, for instance, only ₹2.10 lakh crore out of the ₹9.23 lakh crore written off was recovered. The sectoral mix has flipped. Write-offs in industry fell from ₹1.5 lakh crore in 2018-19 to ₹37,716 crore in 2024-25. Leading the pack is retail loans at ₹45,404 crore, followed by services at ₹38,438 crore, smaller enterprises at ₹28,587 crore and agriculture at ₹21,882 crore.
The industrial NPA crisis did not so much get solved as get replaced by a quieter one. The term ‘write-off’ also gets confused with ‘waiver’. A farm loan waiver is real relief, paid for by the exchequer, extinguishing the farmer’s liability; a bank write-off extinguishes nothing on paper. One is honestly named forgiveness. The other only behaves like it.
After write-off, the bad loan lands at the National Company Law Tribunal. As of June 2026, in 1,480 insolvency cases entailing ₹4.35 lakh crore had been recovered against admitted claims of ₹14.27 lakh crore. The ratio: under 31 percent. The average hides the gap in recovery.
In the Lanco Thermal case, for instance, lenders received just ₹136 crore against claims of ₹33,331 crore—less than 0.5 percent. Small consolation: the Code still beats a fire sale. Nationally, the recovery rate of written-off loans has climbed from 19.14 percent in 2021-22 to 34.77 percent in 2024-25. But an information request from the Central Bank of India revealed shades of recovery—in 10 years, the bank had recovered nearly 74 percent of what it had written off for small and poor borrowers, and just 14.5 percent for large corporate accounts.
The money that stays stuck does so mostly in liquidation, the route the Code meant as a last resort. Of 3,074 cases that ended in liquidation orders, 1,757 have been fully closed, and creditors recovered under 40 percent of the ₹3.44 lakh crore claimed. What’s left apart from lapsed contracts and departed staff are idle factories, worth much less as scrap by the time anyone sells it.
There is a chasm between the Code’s promise and its performance. The law’s design rested on speed: a case admitted within 14 days, resolved within 330. In practice, admission alone now averages roughly 500 days and resolution another 900—3-4 years inside a process built for 11 months.
As of March 2026, 1,885 insolvency cases were pending and nearly three in four of them had already run past that 330-day limit. Delays are an Indian condition. The NCLT’s pending cases docket—insolvency cases alongside oppression, mergers and issues under Companies Act—stood at 14,813. Every one of those months is where a ₹14-lakh-crore claim quietly becomes a ₹2.6-lakh-crore fire-sale.
Missing borrowers—now defined as fugitives—are where this stops being a story about companies and becomes one about people. As per data shared in Parliament, 15 individuals were declared fugitive economic offenders as of October 2025, nine of them linked to large-scale frauds against public-sector banks. The government’s own annexure put the financial loss in these accounts at ₹58,082 crore and recoveries at ₹19,187 crore. Only the Sandesara brothers of Sterling Biotech have settled—not with their banks, but at the Supreme Court under Article 142. The brothers paid up ₹5,100 crore—even as lenders claimed outstanding dues ran to ₹19,283 crore—for all the criminal proceedings against them to be quashed. Nirav Modi has a parade of banks to pay.
Fugitives are only the visible edge. As of June 2025, public-sector banks reported 2,104 ‘wilful defaulters’, with ₹1.77 lakh crore outstanding. The contrast is revealing: the banking system has spent years shrinking the NPA number on its books; the harder question is how much remains attached to the people who borrowed, defaulted and, in some cases, simply refused to pay. And yet cannot be named in public like the fugitives are.
There is one more number the headline must confront: the cost borne by the public. When banks write off loans, the loss is absorbed through provisions and reduced bank income. In the case of public-sector banks, the losses require induction of capital from the government. That makes recovery more than a question of accounting. It is also a question of budget allocation and efficiency of process—which borrowers are pursued, which assets are sold, time consumed by litigation and who bears the unrecovered amount. The cost to Citizen Janardhan is not computed nor shared.
The net NPA ratio tells you the banking system has stopped counting its old losses. Did the money come back to the till? By the government’s own count, that answer is ₹35 out of every ₹100.
Shankkar Aiyar | THE THIRD EYE | Author of The Gated Republic, Aadhaar: A Biometric History of India’s 12 Digit Revolution, and Accidental India
(Views are personal)
(shankkar.aiyar@gmail.com)