Banking Sector Accounting Logic

First published 22 Aug 2026 · Last verified 29 Aug 2026

On the mid-hills road to Surkhet, the branch manager of a small development bank once told a visiting NRB inspection team that his loan book was "clean" — every account current, every file complete. Nineteen months later, that same institution's non-performing loans stood at 40.85 percent of its portfolio, its capital had evaporated, and Nepal Rastra Bank walked in under Section 86(B) of the NRB Act to take the institution's management away from its own board. Nothing about that collapse was sudden. It was written, quarter after quarter, in provisioning tables, interest suspense accounts, and a capital adequacy ratio that kept sliding toward a floor nobody wanted to look at directly. This chapter teaches you to read those tables before the inspection team has to.

Lesson 29.1 — The Architecture of a Nepali Bank's Financial Statements

A Nepali commercial bank's annual report is not built like a manufacturing company's. There is no "cost of goods sold," no inventory turnover, no gross margin in the conventional sense. A bank sells money and buys money, and its entire income statement is a spread business layered on top of a balance sheet that is, by design, almost entirely made of other people's liabilities and claims.

Start with the balance sheet. On the liability side, the dominant line is deposits — current, savings, call, and fixed deposits from the public, disclosed under NRB's prescribed format (per the Directive on Format of Financial Statements) and broken down by institution type, maturity, and currency. Beneath deposits sit borrowings (from NRB's refinance facilities, interbank borrowing, and occasionally bonds or debentures), then "other liabilities," then shareholder equity — share capital, share premium, retained earnings, and a set of statutory and regulatory reserves that exist only because NRB requires them, not because the bank chose to hold them.

On the asset side, "Loans and advances to customers" is the single largest and most consequential line for everything that follows in this chapter. Beside it sit investments (government securities, NRB bonds, corporate debentures, and increasingly, mutual fund units and hybrid instruments), "cash and balances with banks," and fixed and other assets. Loans and advances are always shown net of impairment — the provision you will spend the rest of this chapter learning to interrogate.

The income statement mirrors this structure. Interest income (from loans, investments, and interbank placements) sits at the top, interest expense (paid on deposits and borrowings) is deducted to arrive at net interest income, and this single number carries more weight in a bank's profitability than any other line in the entire statement. Below net interest income comes fee and commission income (loan processing fees, LC/guarantee commissions, remittance fees — collectively "non-funded" income), then net trading and other operating income, then the impairment charge for loans and other assets, then personnel and operating expenses, and finally profit before and after tax.

Understanding this architecture matters because every subsequent lesson in this chapter — provisioning, capital adequacy, income recognition — is really an argument about how honestly the numbers in these two statements represent economic reality. A bank's published net profit is, more than in almost any other sector, a policy choice as much as an observed fact: how aggressively did management classify loans, how much interest income did it recognise on stressed accounts, how conservatively did it provision under NFRS 9 versus the regulatory floor. Learning to read a bank means learning to see the choices behind the numbers, not just the numbers.

Deposit mix deserves a moment on its own, because it drives both funding cost and liquidity risk. Nepali banks disclose the proportion of deposits held as fixed deposits versus current and savings (CASA) accounts. A bank overly reliant on fixed deposits — sector-wide fixed deposits made up roughly 48 percent of total funding as of mid-August 2025 — carries a higher cost of funds and is more exposed to the "differential" wars that break out whenever liquidity tightens, as banks bid up FD rates against each other to retain depositors. A bank with a strong CASA base has cheaper, stickier funding and, all else equal, a structurally higher NIM.

Lesson 29.2 — Interest Income Recognition and the Interest Suspense Account

The single most consequential accounting judgment a Nepali bank makes every quarter is whether to recognise interest income on a loan on an accrual basis or to stop recognising it and route it instead to an "interest suspense" account. Get this wrong — deliberately or through weak systems — and a bank can report profit that does not exist.

NRB's rules on this have tightened materially in recent years. The original Guideline on Recognition of Interest Income, 2019 required banks to stop accruing interest income to the profit and loss account once a loan became non-performing (broadly, once it crossed into Substandard, Doubtful, or Loss classification, i.e., overdue beyond three months) and to instead park the accrued-but-uncollected interest in an interest suspense account on the balance sheet, recognised as income only upon actual cash collection. NRB followed this with a stricter Guidance Note on Interest Income Recognition in 2025, tightening the treatment further — narrowing the circumstances under which banks could continue accruing interest on loans showing early signs of stress (including certain Watchlist-category exposures) and requiring closer alignment between classification status and income recognition.

For an analyst, this creates a specific and very practical check: compare the growth in a bank's reported interest income against the growth in its interest suspense balance (disclosed in the notes to accounts) and against the growth of its non-performing loan book. If NPLs are rising but interest suspense is flat or falling, and interest income is still growing briskly, that is a red flag worth chasing — it suggests either aggressive reclassification delays (loans that should have moved into non-performing categories are being kept "current" through evergreening or restructuring) or genuinely improving asset quality. The notes to accounts, not the headline income statement, are where this distinction gets resolved.

This is also where "loan evergreening" enters the picture — a bank rolling over or refinancing a stressed borrower's facility just before it would otherwise cross a classification threshold, effectively resetting the overdue clock. NRB's Watchlist category, added specifically to catch this behaviour, requires banks to flag loans where the borrower has shown negative cash flows for three consecutive years or where credit quality is deteriorating even without a formal overdue event — closing some, though not all, of the gap that evergreening exploits.

Interest suspense recognition sits beside a second income-statement discipline: the treatment of "staff bonus" under the Bonus Act, which is calculated as a percentage of net profit before bonus and tax and expensed above the tax line — a small but non-trivial reconciling item between operating profit and pre-tax profit that is easy to overlook when comparing bank margins to non-bank companies.

Lesson 29.3 — Loan Classification and Provisioning Mechanics

This is the load-bearing wall of Nepali bank accounting. NRB's Unified Directives to Banks and Financial Institutions — specifically the directive governing loan classification and loss provisioning (commonly referenced as Directive No. 2 in the annually reissued Unified Directives) — set out a five-tier classification system, moving a loan from performing to non-performing purely as a function of days past due, supplemented by qualitative red flags.

The five categories and their minimum provisioning requirements are:

Restructured and rescheduled loans sit in a special sub-bucket, provisioned at rates that vary with the underlying reason for restructuring and the loan's original classification — ranging roughly from 12.5 percent up toward the full provisioning rate applicable to the category the loan would otherwise have fallen into. This matters enormously in practice: a bank under earnings pressure has every incentive to restructure a stressed loan rather than let it migrate into Substandard or Doubtful, because restructuring can — if not scrutinized — reset both the overdue clock and the provisioning burden.

Two structural features of this system are worth internalizing. First, provisioning is set on the gross loan exposure by classification bucket, not on a loan-by-loan discounted cash flow estimate of expected recovery — this is what makes it a "prudential" or "regulatory" provisioning regime rather than an economic-loss regime. Second, the entire loan book (not just the non-performing portion) carries some provisioning — even a fully current, fully performing Pass loan carries a mandatory 1.25 percent general provision. This means "provision coverage" as a concept in Nepal always has a floor that has nothing to do with credit quality; it is baked into the classification system itself.

For the analyst, the practical work is this: pull the loan classification breakdown from the notes to accounts (every NRB-format annual report discloses it), calculate what fraction of the book sits in each bucket, and track the migration between buckets quarter over quarter. A bank whose Watchlist bucket is swelling while Pass shrinks, even if Substandard/Doubtful/Loss look stable, is telling you where next year's non-performing loans are going to come from.

Lesson 29.4 — NFRS 9 Expected Credit Loss and the Regulatory Reserve Bridge

Since the mandatory adoption of Nepal Financial Reporting Standards for banks, Nepali commercial banks have been required to compute impairment under NFRS 9's expected credit loss (ECL) model in their audited financial statements — a fundamentally different logic from the directive-based classification system in Lesson 29.3, running in parallel to it rather than replacing it.

NFRS 9 requires a three-stage approach. Stage 1 covers loans with no significant increase in credit risk since origination; these carry a 12-month ECL — the portion of lifetime expected losses that could occur within the next twelve months. Stage 2 covers loans that have shown a significant increase in credit risk (even if still technically performing) and requires a full lifetime ECL — the expected loss over the entire remaining life of the loan. Stage 3 covers credit-impaired loans (broadly aligned with, though not identical to, the regulatory non-performing categories) and also carries a lifetime ECL, now computed against a loan the bank accepts is impaired.

Crucially, ECL under NFRS 9 is forward-looking and probability-weighted: it incorporates macroeconomic scenarios, historical loss experience, and borrower-specific risk factors, rather than applying a flat percentage by overdue bucket. NRB's NFRS 9 Expected Credit Loss Related Guidelines (issued 2024, subsequently amended) set the operational parameters Nepali banks must follow when building these models — including standardised approaches to probability of default, loss given default, and macroeconomic overlay factors, so that ECL estimates across the sector are not each bank's unconstrained internal judgment.

That mechanism is the Regulatory Reserve. Where a bank's NFRS 9 ECL impairment is lower than the NRB directive-based provision the loan classification would otherwise require, the shortfall must be transferred out of distributable retained earnings into a Regulatory Reserve within equity — annually, as part of the appropriation of profit. This reserve is not available for dividend distribution; it exists purely to prevent a bank from using a more lenient internal ECL model to report and distribute profit that NRB's prudential framework says has not actually been earned yet. If NFRS 9 impairment happens to be higher than the directive minimum in a given period, no such transfer is required — the higher, more conservative NFRS 9 number simply stands.

Because the shift to full NFRS 9 ECL had the potential to create a sudden capital shock for banks with previously under-provisioned books, NRB built in a transitional arrangement: a four-year adjustment window (spanning fiscal years 2081/82 through 2084/85) during which banks receive CET1 capital relief against the Day 1 impact of ECL adoption, declining from roughly 80 percent relief in the first year down to 20 percent by the final year, before full ECL impact flows through to capital unmitigated. This transitional relief is itself disclosed in the capital adequacy notes, and an analyst should check whether a bank's reported CAR is still benefiting from this glide path — a bank that looks adequately capitalised today, with the benefit of transitional relief, may look considerably tighter once the relief fully expires.

Lesson 29.5 — Capital Adequacy: The Ratio That Decides Whether a Bank Survives

Capital adequacy is the mechanism by which NRB ensures a bank has enough of its own shareholders' money at risk, relative to the riskiness of its assets, to absorb losses before depositors and the deposit-guarantee system are ever called upon. It is governed by a separate strand of the Unified Directives — the New Capital Adequacy Framework, aligned broadly with Basel III as adapted for Nepal's banking system — and it is the single ratio NRB watches most closely as a trigger for supervisory intervention.

The framework requires commercial banks to maintain a minimum Capital Adequacy Ratio (CAR) — total qualifying capital divided by risk-weighted assets — of 11 percent, of which minimum core capital (Tier 1, essentially paid-up equity capital, share premium, and retained earnings, net of specified deductions) must be at least 8.5 percent. The balance between the 8.5 percent Tier 1 floor and the 11 percent total floor can be met with supplementary (Tier 2) capital — general loan loss provisions up to a specified limit, subordinated debt instruments, and revaluation reserves, among other qualifying items. Beyond these baseline requirements, the framework layers on a capital conservation buffer and, for banks deemed systemically important, additional loss-absorbency requirements — meaning the "true" minimum a well-run, systemically significant bank should be targeting in practice typically runs above the bare regulatory floor.

To see that dispersion concretely, look at how individual banks have actually reported against these floors. In one NRB-compliance snapshot (Chaitra-end 2078), Standard Chartered Bank Nepal reported the highest CAR in the sector at 15.90 percent, while Himalayan Bank Limited reported the lowest compliant figure at 11.61 percent — barely above the 11 percent floor. On core capital, NIC Asia Bank and Prabhu Bank both reported Tier 1 ratios in the 8.54–8.70 percent range — again, only marginally above the 8.5 percent minimum. All 27 commercial banks operating at the time were technically compliant, but "technically compliant" and "comfortably capitalised" are not the same claim, and an analyst who stops at the pass/fail line misses the more useful information: how much room a bank actually has before the next credit cycle pushes it toward the floor.

Why does this matter so much in practice? Because CAR breach is one of the principal triggers for NRB's Prompt Corrective Action (PCA) framework — a graduated set of restrictions (on dividend payment, branch expansion, senior management changes, and eventually direct intervention) that NRB applies to banks and financial institutions falling below capital, asset-quality, or governance thresholds. Karnali Development Bank was placed under PCA in November 2024 after failing to maintain its required capital adequacy ratio; when that measure proved insufficient against the bank's deteriorating position — non-performing loans that had by then reached 40.85 percent of its portfolio, compounded by weak institutional governance and a liquidity crunch that left it unable to meet deposit repayment obligations — NRB moved a step further in December 2024, assuming direct management control under Section 86(B) of the Nepal Rastra Bank Act, 2002, appointing an NRB-led management team to run the institution, protect depositor funds, recover loans, audit assets and liabilities, and pursue accountability for the underlying financial misconduct.

At the other end of the spectrum, sector-wide non-performing loans stood at roughly 4.62 percent as of mid-August 2025, with sector-wide loan loss provisions running about 5.09 percent of total loans — figures that look moderate in isolation but that NRB and market commentary have flagged as trending upward against a backdrop of high credit-to-GDP exposure (above 91 percent) and heavy reliance on fixed-deposit funding. None of this, on its own, signals crisis. It signals exactly what this chapter is about: numbers that must be read in context, in trend, and against the regulatory floor, rather than as a single static "pass" or "fail."

Lesson 29.6 — Reading a Bank's Numbers Like an Examiner: A Practical Checklist

Everything in this chapter converges on a discipline: reading a Nepali bank's financial statements the way an NRB supervisor reads them, not the way a casual investor skims a headline EPS number. Use the following sequence every time you pick up a commercial bank's quarterly or annual disclosure.

Start with capital. Pull the CAR and Tier 1 ratio, and measure the distance to the 11 percent and 8.5 percent floors respectively, not just whether the bank clears them. Check the capital notes for NFRS 9 transitional relief and note whether the current ratio depends on a glide path that is shrinking year by year.

Move to asset quality. Extract the full loan classification breakdown — Pass, Watchlist, Substandard, Doubtful, Loss, and Restructured/Rescheduled — as rupee amounts, not just the summary NPL ratio. Build (or update) your migration matrix across at least four quarters. A rising Watchlist or Restructured bucket, even alongside a flat headline NPL ratio, is your earliest warning signal.

Check provisioning discipline next. Compare directive-based provisioning against NFRS 9 ECL impairment in the notes to accounts, and track the Regulatory Reserve movement in the statement of changes in equity. A growing Regulatory Reserve alongside strong reported profit tells you a meaningful share of that profit is not yet available to shareholders.

Then examine income quality. Separate net interest income (and NIM) from fee/commission income and any trading or one-off gains. Cross-check interest income growth against interest suspense account growth and NPL growth in the notes — divergence between these three is a signal worth investigating before it shows up in the headline numbers.

Finally, situate the bank within its sector context: compare its CAR, NPL ratio, and NIM against sector averages (roughly 13 percent CAR, 4.6 percent NPL, and provisioning near 5 percent of loans as of the most recent NRB-referenced figures), and ask whether the bank is an outlier in a direction that matters — a bank meaningfully below sector CAR or meaningfully above sector NPL deserves closer scrutiny than one that tracks the average.