Where India’s next stressed assets are forming: non-bank loan books, stalled housing projects and renewable energy plants
India’s first cycle of stressed assets was dominated by large corporate loans held by banks. The stress forming now is smaller in ticket size, more widely dispersed and more often held outside the banking system, and each of the sectors producing it needs a different capability to resolve.
Contents
How a borrower becomes distressed
Default follows identifiable patterns. Most cases trace to excessive leverage, operational decline or a failure of governance, and many combine more than one. Each pathway leaves a different asset behind and calls for a different route to recovery, so a portfolio priced as though every default were alike will be mispriced.
Over-leverage is the most common pathway and the most tractable. The asset keeps its economic value; what has failed is the capital structure. In an illustrative composite, a developer in Pune finances a 400-unit residential tower with a ₹200 crore term loan, expecting pre-sales to fund the rest of construction. The market softens, pre-sales slow and construction stops at 60 per cent, with 200 units already sold and deposits collected. The loan becomes non-performing, but the land and the structure remain, and the finished project would be worth well above the debt. Recovery comes from acquiring the claim at a discount, restructuring the capital stack and completing the building.
Operational decline is harder to resolve. In a second illustrative composite, a textile company in Gujarat installs modern looms in 2012 with ₹150 crore of bank debt, on a plan built around exports to Europe. By 2016 competition from Bangladesh and Vietnam has eroded its pricing and domestic demand cannot fill the capacity, and the company defaults. The recovery question extends beyond the collateral to whether the business, or a reconfigured version of it, can be made profitable again: whether the plant can be repurposed, whether costs can be brought into line with available revenue, and whether the downturn is cyclical or structural. Those answers need sector operating knowledge.
Governance failure runs from poor judgment, such as expansion into unfamiliar sectors and overpriced acquisitions, to fraud: fictitious invoices, inflated receivables and loan proceeds diverted into personal property. Governance cases trade at the deepest discounts, because credit impairment comes with reputational damage that keeps other buyers away and makes lenders willing to sell at almost any price. They also carry the highest execution risk, including disputed title to collateral, asset stripping and hostile litigation from the former promoter.
What the first cycle settled, and how the current one differs
India’s first institutional cycle of stressed investing followed the recognition of bank non-performing assets (NPAs) after the Reserve Bank of India’s 2015 Asset Quality Review (see “How large India’s stressed debt market is, and who holds it now”). It was concentrated in large corporate loans to infrastructure, power and metals, mostly originated between 2008 and 2014 and held by public sector banks. Large cases under the Insolvency and Bankruptcy Code, several of them in steel, dominated the cycle, drew global capital and showed that the Code could deliver recoveries.
The cycle absorbed the largest and most visible pool of legacy bank loans, and headline NPA ratios have since fallen well below their peak. Reading that as a sign that the opportunity is narrowing confuses the end of one cohort of large corporate defaults with the end of stress, which a large and fast-growing credit market continues to generate.
The cycle’s more durable legacy is legal. The largest cases, with claims running to tens of thousands of crore and litigation through the National Company Law Appellate Tribunal and the Supreme Court, established that the commercial judgment of the committee of creditors receives minimal judicial interference, that dissenting financial creditors can receive less than their proportionate share where the committee so decides, and that an approved plan gives the acquirer a clean slate. Later cases clarified the treatment of personal guarantors and the priority of secured over unsecured creditors (see “India’s insolvency process step by step: admission, resolution plans and settlement”). Recovery estimates are more reliable today than in 2017 or 2018 as a result.
The current cycle differs in character. It is more granular, more of it sits outside the banking system, and three sectors account for most of it. Resolving a single large corporate account is mainly a matter of law and negotiation, conducted before tribunals and within a committee of creditors. The stress forming now is more often an operational problem, and each of the three sectors needs a different capability.
| Sector | Nature of the problem | Main recovery route | Capability required |
|---|---|---|---|
| Non-bank and housing finance loan books | Thousands of small secured and unsecured loans spread across states | Settlement first, then enforcement of security and auction | Field servicing and collection, enforcement in several states, account-level data systems |
| Stalled housing projects | A sound asset under a failed capital structure | Possession, completion under state real estate regulation and sale of finished units | Construction management, regulatory approvals, local sales knowledge |
| Renewable energy plants | Operating plants with a cash shortfall | Operational turnaround and resizing of project debt | Plant operations, negotiation with state distribution companies, warranty claims |
A buyer equipped only for large corporate resolution is equipped for the previous cycle; one that can also service granular books can shift its weighting between the two as relative pricing moves. Figures below that describe typical pricing, costs and practice in these sectors are as at April 2026 unless another date is given.
Non-bank and housing finance loan books
Between 2015 and 2019, non-banking financial companies (NBFCs) grew their balance sheets at compound rates above 15 per cent a year. Housing finance companies were particularly aggressive, lending to borrowers banks had underserved: the self-employed, earners with informal incomes and buyers in Tier 2 and Tier 3 cities. Funding came from bank lines, commercial paper and bonds, and that asset-liability mismatch proved damaging when the default of a large infrastructure lender in September 2018 set off a liquidity squeeze across the sector.
Risk had moved to lenders under a lighter capital regime, and the resulting stress shows most clearly in small-ticket lending. Stressed assets in NBFC microfinance rose to 5.9 per cent at March 2025 from 3.9 per cent six months earlier, and write-off ratios in the sector reached levels well above their pre-2022 range. In personal loans below ₹50,000, NBFCs, housing finance companies and fintech lenders hold the overwhelming majority of exposure, and a large share of those borrowers already had three or more active loans when the new loan was made, the clearest available sign of layered borrowing.
3.9% to 5.9%
Unsecured microfinance and personal loans carry little collateral and are recovered mainly through settlement and behavioural collection, with borrower data quality often the main diligence problem. Secured NBFC and housing finance books are the part of the sector most often sold as stressed portfolios.
What a secured portfolio looks like
Loans in a secured portfolio are small, typically ₹15 lakh to ₹5 crore, and secured by residential and commercial property in identifiable locations. The borrowers are mostly individuals and small businesses, so enforcement, whether under the Securitisation and Reconstruction of Financial Assets and Enforcement of Security Interest Act, 2002 (SARFAESI) or through settlement, is simpler than in a corporate group restructuring. A seller under liquidity pressure from its own creditors accepts a deeper discount than a healthy bank, and prices of 25 to 45 per cent of outstanding are typical for such sales, deeper than the segment-wide range (see “How large India’s stressed debt market is, and who holds it now”).
Granularity also changes how risk behaves. Across five hundred mortgage loans in three states, some borrowers settle quickly, others need enforcement and a few are written off, but the blended recovery clusters in a fairly narrow band because the positions are small and independent. A book concentrated in one region or one borrower type loses much of that averaging.
How a granular book is worked, and what it needs
Recovery on a granular portfolio is intensive but methodical. A field team, typically twenty to forty people across the relevant states, splits into settlement negotiators and enforcement specialists. Borrowers who can raise funds, from another asset, family or a new lender, are typically offered 20 to 40 per cent off the outstanding balance in exchange for a single immediate payment. Those who cannot or will not settle receive SARFAESI demand notices, followed by possession through the District Magistrate and auction of the collateral.
Recovery follows a predictable sequence over time:
- Months one to twelve produce most of the negotiated settlements, typically 35 to 45 per cent of total recovery.
- Months twelve to thirty produce the core enforcement recoveries, as possession proceedings and auctions are completed across jurisdictions.
- Months thirty to forty-eight resolve the tail: hostile borrowers, contested titles and properties that need court-assisted possession.
- Typically 5 to 10 per cent of accounts are written off entirely, because the collateral is too impaired or too remote to justify further enforcement cost.
Outcomes vary mainly with collateral quality and the efficiency of the enforcement jurisdictions, both assessable before purchase (see “Enforcing a secured loan under SARFAESI: from demand notice to auction”), and with borrower cooperation, which profiling predicts only in part.
The capability that decides the outcome is servicing and collection at scale: field presence across several states, trained negotiators, enforcement staff who can carry a possession through, and systems that track thousands of accounts. Litigation expertise matters less here than in a large corporate case. Buying a granular portfolio without that capacity is a common and expensive mistake, and a buyer that relies on the seller’s servicer depends on an organisation whose incentives are rarely aligned with its own once the sale has closed.
Stalled housing projects
Mid-market developers in Tier 1 and Tier 2 cities, with projects of ₹50 to 300 crore, show a recurring pattern. A project launched between 2014 and 2017 was financed by a term loan sized on the assumption that pre-sales would fund most of the construction, and typically reached 50 to 70 per cent completion. Then conditions turned. The Real Estate (Regulation and Development) Act, 2016 (RERA) added compliance costs, demonetisation disrupted the cash-heavy parts of the market, and the Goods and Services Tax changed the tax economics of under-construction property. Pre-sales slowed, work stopped, interest accumulated and the account became non-performing.
The assets behind these projects have real, assessable value: land in identifiable locations, partly completed structures with quantifiable completion costs, and receivables from units already sold. A holder with capital and project management capability can often finish what the developer could not. It acquires the debt, takes possession under SARFAESI, appoints a third-party project manager and funds the remaining construction, at a cost typically 20 to 35 per cent of the eventual sale value of the completed inventory. The margin of safety comes largely from the land, which retains its value even if finished units sell below projection.
RERA has improved the conditions for this kind of recovery. It requires project registration, mandates that project funds be held in dedicated escrow accounts and provides a framework for managing existing buyers’ expectations during a change of developer, and several state RERA authorities have developed procedures that recognise completion led by an asset reconstruction company. SARFAESI provides the means of taking possession and RERA the framework for completing and selling.
The main risks are construction cost overruns, delays in regulatory approvals and slow absorption of finished units in the local market. Buyers who paid deposits before the project stalled must be accommodated in any completion plan, and their remaining payments on handover form part of the recovery (see “A stressed loan portfolio from purchase to recovery: an illustrative walk-through”).
Renewable energy plants
India’s installed solar and wind capacity exceeded 200 gigawatts as at April 2026, financed largely through project-level special purpose vehicles carrying term debt. Many were built at tariffs bid in competitive auctions, where bidders assumed falling panel costs, reliable generation and prompt payment by state electricity distribution companies (DISCOMs). In some cases construction costs exceeded projections, generation fell short where panels degraded faster than their warranties allowed, and DISCOMs delayed payments by six to twelve months, opening cash gaps the project’s capital structure could not absorb.
Distress of this kind is mainly operational. The assets are working plants with long remaining lives and contracted revenue under power purchase agreements, and recovery depends on running them well, with little reliance on court outcomes or auction buyers.
An illustrative composite: four 25 megawatt solar plants in Rajasthan carry aggregate project debt of ₹180 crore. Revenue runs 30 per cent below projection because the state DISCOM is paying late and panel degradation has exceeded specification, and the sponsor cannot fund the gap in debt service. A buyer acquires the debt at 40 to 45 per cent of outstanding, ₹72 to 81 crore, and then works through four steps:
- replace the operations and maintenance contractor with a lower-cost provider, cutting annual operating costs by 15 to 20 per cent;
- convert DISCOM arrears into a staggered payment plan;
- file warranty claims against the panel manufacturer;
- restructure the project debt at a lower interest rate that reflects the reduced risk.
Within twelve to eighteen months revenue is back in line with the original base case, and the debt service coverage ratio, operating cash flow divided by debt service, is stable above 1.2; below 1.0 a project cannot service its debt from its own cash. With a twenty-year power purchase agreement behind it, the plant then produces cash flows more stable than real estate sale proceeds, and the execution risk is lower than in enforcement.
Renewable plants are likely to become a larger source of stressed assets over the coming decade. Capacity keeps expanding on project debt, and with long gestation, uncertain tariff bidding, DISCOM payment delays and changing technology, some share of every cohort of projects will run into difficulty.
How the credit cycle sets supply
India’s total credit stock across banks and NBFCs exceeded ₹200 lakh crore as at April 2026. Across Indian banking cycles, new NPAs have formed at an average of approximately 3 to 5 per cent of outstanding credit a year: below 2 per cent in benign periods, above 5 per cent in periods of stress, and occasionally above 8 per cent in severe downturns, as in the recognition cycle of 2015 to 2018.
At a steady formation rate of 3 per cent on the current stock, roughly ₹6 lakh crore of new non-performing assets arise each year. If half are resolved between borrower and original lender and never enter the stressed debt market, about ₹3 lakh crore a year becomes available for acquisition by asset reconstruction companies and funds. Because most fresh defaults in any year come from loans already on the books, this stock-based measure is a better guide to annual supply than the narrower annual addition from new bank lending, which serves as a floor (see “How large India’s stressed debt market is, and who holds it now”).
The cycle adds variation around that baseline, in recognisable phases:
- Expansion. Credit grows quickly, underwriting loosens and leverage rises. For banks this ran from 2010 to 2014; for NBFCs, from 2015 to 2019.
- Recognition and resolution. The stressed pool expands sharply and acquisition pricing widens. For banks this ran from 2015 to 2021; for NBFCs it is the current period.
- Normalisation. The pool contracts as assets are resolved and new lending is more disciplined, while enforcement is faster because the tribunals and the SARFAESI machinery are less congested.
Even in normalisation the pool does not disappear, because the baseline formation rate keeps adding to it. The lag between origination and recognition has typically been four to seven years in Indian cycles: lending of the late 2000s surfaced as recognised stress from 2015 onward. That lag makes the supply of stressed assets unusually forecastable.
The next expansion
Projections current in January 2026 put Indian corporate credit at roughly 57 per cent of gross domestic product, rising to 69 per cent by 2028. A rise of that size, after a decade of broadly flat corporate leverage and with a cleaned-up banking system whose appetite for risk has returned, sets the conditions for the next cohort of stress. This is no forecast of crisis, since credit expansion in a growing economy is ordinary and largely productive. Some proportion of any expansion will fail to perform, and on past lags it will reach the market several years after the lending.
Prepared for educational purposes only. Not an offer to sell securities, investment advice or a recommendation. Views are as at the date of publication and are subject to change.