Asset Classification, Provisioning and Non-Performing Loans
Nepali banks provision loans on a five-step ageing ladder, and an independent review found the reported bad-debt ratio too low.
The five categories
Loan classification in Nepal is mechanical and ageing-driven. Directive 2 of the Unified Directives sorts every credit exposure into one of five categories by how long it has been overdue, and attaches a minimum provision to each.
| Category | Overdue period | Minimum provision | Status |
|---|---|---|---|
| Pass | Not overdue, or up to 1 month | 1% | Performing |
| Watchlist | 1 to 3 months | 5% | Performing |
| Substandard | 3 to 6 months | 25% | Non-performing |
| Doubtful | 6 months to 1 year | 50% | Non-performing |
| Loss | Over 1 year | 100% | Non-performing |
The headline NPL ratio published by NRB and by every Nepali bank is the sum of the last three categories over total loans. Watchlist is excluded.
Two features of this table deserve more attention than they usually get.
The Watchlist provision is severe. Five per cent against a loan that is between one and three months late, and which is still counted as performing, is high by regional standards. The migration of a large exposure from Pass to Watchlist costs a Nepali bank five times the provision at a point where most systems would charge nothing beyond a general reserve. The commercial consequence is a strong institutional incentive to keep borrowers out of Watchlist, by restructuring, by refinancing, or by capitalising interest. That incentive is the mechanism behind the evergreening the loan portfolio review identified.
The bucket boundaries are calendar boundaries. Nepal’s fiscal year ends in mid-July, and quarterly reporting drives classification. A borrower who pays on the last working day of Ashadh is Pass; the same borrower paying a week later is Watchlist. Anyone reading a Nepali bank’s quarterly numbers should expect the fourth quarter to look better than the underlying position.
Provisions can be reversed when a loan is upgraded, and NRB has periodically moved the pass-loan rate as a counter-cyclical lever. It stood at 1.25% historically, was set at 1.2%, and was reduced to 1.0% for FY 2082/83 (2025/26). Because general provisions on performing loans count toward Tier 2 capital up to 1.25% of credit risk-weighted exposures, that adjustment moves reported capital as well as reported profit.
Exceptions to the ageing test
Several rules modify the raw ageing test, and the December 2025 amendment round changed most of them.
- Loans under auction or recovery proceedings attract 100% provision, which may be discontinued if the loan is regularised after recovery action began.
- Collateral taken from a defaulter may be recognised as a non-banking asset only after three separate auctions have failed to sell it. Before that change, the asset came onto the balance sheet earlier and the loan came off.
- Restructuring where only the interest rate is adjusted no longer requires additional provision, once a year, at the borrower’s request and subject to an income assessment.
- Restructured loans in nationally prioritised sectors that are financially stressed because of capacity expansion continue at 1%.
- Loans above NPR 2 billion extended without a consortium agreement are classified to Watchlist, a rule that ties classification to the exposure limits discussed in Exposure Limits: Single Obligor, Group and Sector Concentration.
The last of these is the clearest example of NRB using the classification ladder to enforce an unrelated rule. The penalty for lending large without a consortium is a 5% provision rather than a fine.
The reported numbers
| Date | System NPL ratio |
|---|---|
| Mid-July 2023 | 3.02% |
| Mid-July 2024 | 3.86% |
| Mid-July 2025 | 4.62% |
| January 2026 (IMF) | 5.4% |
| Mid-February 2026 | 5.42% |
| Mid-April 2026 | 5.60% |
At mid-July 2025 total non-performing loans were NPR 258.19 billion, against NPR 199.66 billion a year earlier. The composition is unusual. Some 62.30% of NPLs sat in the Loss category and 21.06% in Doubtful, so the great majority of Nepal’s bad debt was written down to zero long ago and is still on the books. That reflects the slowness of collateral realisation rather than the pace of new default.
Sectoral concentrations at mid-July 2025 ran highest in fishery (9.97%), wholesale and retail trade (8.51%) and construction (8.40%). By mid-April 2026 agriculture accounted for 23.5% of all non-performing loans and microfinance institutions were reporting an 11.32% NPL ratio against roughly 7% a year earlier, with NPR 54.08 billion of bad loans. Directed lending and asset quality are connected, and the connection is examined in Exposure Limits: Single Obligor, Group and Sector Concentration.
Total loan loss provision across the system stood at 5.92% of total loans at mid-April 2026, marginally above the 5.60% NPL ratio.
The loan portfolio review
NRB commissioned an independent loan portfolio review of the ten largest commercial banks as a structural benchmark under Nepal’s Extended Credit Facility arrangement with the IMF. The onsite work was a benchmark for end-May 2025 and ran late. The board approved the report in May 2026.
The IMF’s own account is careful and unambiguous in direction. The review indicated NPL levels higher than reported, applied both existing NRB requirements and international practice, and identified reclassification and additional provisioning needs. Nepali press reporting of the report put the actual ratio at 7.7% against 5.5% reported at mid-April 2025, with three banks above 10%, widespread evergreening, and two banks below capital requirement. The report itself is not public, so those specific figures rest on press accounts rather than on NRB or IMF publication.
Two qualitative findings from the earlier 2024 round are worth carrying forward, because they describe the mechanism rather than the magnitude. Banks had not consistently applied the classification and provisioning rules. And loan files frequently lacked project documentation, monitoring records and site visit records. A classification system that depends on the bank identifying overdue status correctly is only as good as the file.
Banks were required to submit NPL Resolution Strategies alongside Capital Restoration Plans by mid-August 2026.
Expected credit loss under NFRS 9
NRB issued guidelines on expected credit loss under NFRS 9 in May 2024, after deferring the project from 2078/79 (2021/22). The model is forward-looking: probability of default, loss given default and exposure at default, with lifetime losses recognised on a significant increase in credit risk, triggered by indicators including 30 days past due or a doubling of lifetime probability of default.
The catch is the backstop. For at least five years, a bank must book the higher of its NFRS 9 expected credit loss and the existing regulatory provision. Since the regulatory ladder is aggressive at the front end, charging 1% on every performing loan and 5% on Watchlist, the regulatory number will usually be the higher one for a book that is not visibly deteriorating. The day-one capital impact is phased into CET1 over five years, with 85% of it added back in year one.
Nepal has therefore adopted the accounting standard without yet changing the number that drives capital. Separately, the alignment of the probation period for non-forborne non-performing loans with the Basel Committee standard was completed at the end of December 2025, and interest income recognition moved from the 2024 ECL guidelines to a 2025 guidance note.
Reading a Nepali NPL ratio
Three adjustments make cross-border comparison honest.
The ratio excludes Watchlist, so a bank with a thin Loss book and a swollen Watchlist book looks healthier than one with the reverse. Watchlist balances are disclosed and should be added when comparing to systems that treat 30-day arrears as impaired.
The ratio is a stock, and in Nepal the stock is old. With more than three fifths of non-performing loans in the fully provided Loss category, the ratio moves slowly in both directions and tells you relatively little about current-year credit formation.
And the ratio is bank-reported. The loan portfolio review is the first independent test of that reporting at scale, and it did not confirm it.
Sources
-
2082 Chait (Mid-April 2026) Banking and Financial Statistics (Monthly)
Nepal Rastra Bank · Data to mid-April 2026; published May 2026 · Tier 2 — Official publication
Supports: A non-performing loan ratio of 5.60% of total loans and total loan loss provision of 5.92% of total loans at mid-April 2026, with system deposits of NPR 7,918.3 billion and credit of NPR 5,906.4 billion.
Accessed 26 August 2026
-
Financial Stability Report, Issue No. 17
Nepal Rastra Bank · Published July 2026; data to mid-July 2025 · Tier 2 — Official publication
Supports: The NPL ratio of 4.62% at mid-July 2025 against 3.86% a year earlier, total non-performing loans of NPR 258.19 billion, the composition of NPLs (62.30% loss, 21.06% doubtful), and sectoral NPL concentrations in fishery, wholesale and retail, and construction.
Accessed 26 August 2026
-
Guidelines on Expected Credit Loss under NFRS 9 — final report
Nepal Rastra Bank, Banks and Financial Institutions Regulation Department · May 2024 · Tier 1 — Primary legal instrument
Supports: The move from the incurred-loss model to expected credit loss, the requirement to book the higher of ECL and regulatory provision for a minimum of five years, the five-year CET1 transitional phase-in, and the deferral of implementation from 2078/79.
Accessed 26 August 2026
-
International Monetary Fund · 8 June 2026 · Tier 3 — Multilateral institution
Supports: NPLs at 5.4% in January 2026; that the loan portfolio review indicated NPL levels higher than reported and identified reclassification and additional provisioning needs; the alignment of the probation period for non-forborne NPLs with the BCBS standard, completed end-December 2025; and the mid-August 2026 deadline for NPL Resolution Strategies.
Accessed 26 August 2026
-
Nepal Rastra Bank · 11 July 2025 · Tier 1 — Primary legal instrument
Supports: The loan loss provision on pass loans maintained at 1.0%, reduced from 1.2% in the prior year.
Accessed 26 August 2026
-
Flash Alert on Circular No. 08/082/83
PKF T R Upadhya and Co. · Circular dated 15 December 2025 · Tier 6 — Professional or news source
Supports: 100% loan loss provision on loans under auction or recovery proceedings; recognition of collateral as a non-banking asset only after three failed auctions; the treatment of restructuring where only the interest rate is adjusted; and the move of interest income recognition from the 2024 ECL guidelines to a 2025 guidance note.
Accessed 26 August 2026
-
Priority lending is dragging down Nepali banks into huge bad debt
The Kathmandu Post · 3 July 2026 · Tier 6 — Professional or news source
Supports: System NPLs of 3.02% at mid-July 2023, 5.42% at mid-February 2026 and 5.6% in the third quarter of FY 2025/26; microfinance NPLs of 11.32% at mid-April 2026 against about 7% a year earlier; and NPR 54.08 billion of microfinance bad loans.
Accessed 26 August 2026
-
NRB approves loan portfolio review of 10 major banks
Clickmandu · 15 May 2026 · Tier 6 — Professional or news source
Supports: That the NRB board approved the loan portfolio review report; that it was conducted by Howladar Yunus and Co. under the IMF Extended Credit Facility; and the reported finding of a 7.7% actual NPL ratio against 5.5% reported as at mid-April 2025, with evergreening identified and two banks below capital requirement.
Accessed 26 August 2026
Related guides and next steps