Case Study 7 — A Bank Failure or Near-Failure

First published 26 Aug 2026 · Last verified 29 Aug 2026

Case Study 7 — A Bank Failure or Near-Failure

Bikash Rai had a rule he was proud of: never buy a stock his mother's cousin sold him at a wedding. But in the spring of 2081 (2024 in the calendar most of the world uses), that is almost exactly how he found Karnali Development Bank Limited, ticker KRBL on NEPSE. His cousin-in-law worked a small hardware shop in Surkhet, banked with KRBL because it was the closest branch to his shop, and mentioned over dal-bhat that the bank's stock was cheap, paid a dividend most years, and "not one of those overpriced Kathmandu banks." Bikash, who by then had been running every stock he touched through the Canon Score for two years, wrote the name in his notebook instead of his brokerage app. That single habit — write it down, don't buy it — is the whole subject of this chapter.

This case study is different from the ones before it. It is not about a company that quietly disappointed shareholders over a few bad quarters. It is about a licensed, deposit-taking financial institution that Nepal Rastra Bank (NRB, the central bank and the regulator of every bank and finance company in the country) formally declared to be in crisis, and then took over. Depositors' money was on the line. Shareholders' capital was wiped toward zero. Criminal investigations followed. This is the chapter where the Canon Score framework gets tested against the worst thing that can happen to a financial institution — and where we ask, honestly, whether following the framework would have kept an ordinary investor out of the wreckage in time.

Lesson 89.1 — The Bank Nobody Was Watching

Start with what Karnali Development Bank actually was, because most Kathmandu-based investors had never heard of it before December 2024, and that fact is itself the first lesson.

KRBL was a "Class B" institution under NRB's four-tier licensing system. In Nepal, banks and financial institutions (BFIs) are sorted into four classes: Class A are commercial banks (the big, well-known names with branches across the country and the deepest capital bases). Class B are development banks, smaller than commercial banks, often built around a particular region, historically created to bring formal banking to areas commercial banks were slow to reach. Class C are finance companies, smaller still. Class D are microfinance institutions, focused on small loans to low-income borrowers, often in rural areas. Karnali Development Bank sat in Class B, built around Nepal's mid-western Karnali region, one of the least commercially developed parts of the country. Its paid-up capital — the money shareholders had actually put in, the base a bank is required to hold before NRB will let it operate — stood at roughly Rs 502.8 million. Its deposits, meaning the money ordinary savers had placed with it, stood at about Rs 5.21 billion. Its loan book, the total it had lent out, stood at about Rs 3.81 billion. On paper, at the end of the first quarter of the 2024/25 fiscal year, it had posted a modest net loss of about Rs 19.8 million and reported a non-performing loan (NPL) ratio — the share of its loans that borrowers were not repaying on schedule — of 7.27 percent. A little weak, an analyst might have said, but not obviously a bank in crisis.

That "on paper" phrase is going to do a lot of work in this chapter.

Why would a retail investor like Bikash even consider a bank this small and this obscure? For the same reasons small development banks attract retail money across NEPSE every year. First, book value: a bank with Rs 502.8 million in paid-up capital and a stated net worth per share often trades at a price close to or even below its book value, which looks, on the surface, like a bargain compared to the richly priced commercial banks. Second, dividend habit: development banks with a long run of paying bonus shares or cash dividends build a reputation among retail investors as reliable income stocks, regardless of whether the underlying loan book actually supports that reputation. Third, and most important for this case, information scarcity: almost no brokerage research desk in Kathmandu covers a Class B development bank based in the Karnali region. There is no analyst call, no coverage note, no institutional investor watching the loan book line by line. The only people who really know what is happening inside the bank are the people running it — and, as this case shows, that can be a very dangerous information gap.

This is the first governance lesson of the case study, and it comes before we even open the financial statements: obscurity is not the same as safety, and it is not the same as value either. A stock that nobody is watching can be genuinely undervalued. It can also be a stock where nobody would notice if the numbers were wrong. The Canon Score framework does not treat "under-covered" as a red flag by itself — plenty of good small companies in Nepal are under-covered simply because research capacity is limited. But it does insist that when coverage is thin, the investor's own homework has to be thicker, not thinner, to compensate. Bikash's mistake would have been treating his cousin-in-law's tip as a substitute for that homework rather than a reason to start it.

Lesson 89.2 — Reading the Governance Red Flags Before the Collapse

Chapters 19 through 23 built the governance side of the Canon Score: who sits on the board, how much power is concentrated in one family or one promoter group, whether related-party transactions are disclosed and priced fairly, whether the audit function is independent, and whether the company communicates with shareholders honestly and on time. Applied to a bank, these questions carry extra weight, because a bank's "product" is trust with other people's money, and a governance failure at a bank does not just cost shareholders — it threatens depositors, and through the deposit insurance and financial stability system, it can touch the whole sector.

Here is what the governance record around Karnali Development Bank looked like, separating what was genuinely visible before the takeover from what investigators only established afterward — a distinction this chapter will return to in Lesson 89.6, because conflating the two is its own kind of hindsight error. What was visible in real time, to anyone reading board composition and AGM records before December 2024: Rajendra Bir Raya, the bank's founding promoter, held both the chairman and chief executive roles from roughly 2019 to 2022 — an unusual concentration of power at a small development bank in itself — before a subsequent chairman, Pashupati Dayal Mishra, a sitting member of a major political party (CPN-UML), took over. Two changes of chairman-level leadership in a bank's first several years, one of them combining the chairman and CEO roles in a single person, is a real, contemporaneously checkable governance flag — the kind Chapters 19 through 23 ask an investor to notice without needing to know what would come next. What was only established later, after NRB's takeover, is the criminal case: in June 2025, roughly six months after the takeover, the Central Investigation Bureau arrested Raya at Belahiya, near the Nepal–India border in Rupandehi district, as he attempted to cross out of the country — not an extradition from abroad, but a domestic arrest of someone who had gone underground inside Nepal after the investigation began. Mishra and several former executives, including a former finance chief, were separately investigated and, in some cases, later remanded in judicial custody in connection with an alleged embezzlement exceeding Rs 3 billion. This is not a story of one rogue employee. It is a story of a chain of leadership, across multiple chairmanships, that investigators allege was connected to the same pattern of financial irregularity — but the criminal confirmation of that pattern came in 2025, well after the stock had already been suspended, and it is important not to credit a 2024 decision with foresight it could not have had into a 2025 arrest.

The Canon Score governance checklist (from Chapter 20 onward) asks a specific, almost boring set of questions that turn out to matter enormously in hindsight. Has the chairman changed hands under unclear circumstances? Has the external auditor changed unusually often, or does the auditor's report contain qualified language that management explains away in a single sentence in the annual report? Are related-party loans — loans to companies or individuals connected to board members or senior management — disclosed with enough detail to check whether they were priced and approved the way an arm's-length loan would be? Is the annual general meeting (AGM) held on schedule, or does it keep slipping? Do quarterly disclosures arrive on time, in full, and readable, or are they thin, delayed, and full of boilerplate?

None of these questions requires access to inside information. They require reading what a company is already required to publish, and noticing what is missing or delayed. For Karnali Development Bank, the retrospective picture that emerged after NRB's takeover — repeated leadership turnover at the chairman level, later-confirmed financial irregularities reaching into the billions of rupees, and a regulatory relationship that (as later criminal proceedings would allege) had itself been compromised — is exactly the pattern the governance chapters warn readers to price in as risk, well before any of it is confirmed as fraud. A disciplined investor does not need proof of fraud to downgrade a governance score. Absence of the normal signals of good governance is itself the signal.

That last point deserves its own sentence, because it is easy to misread. Nepal Rastra Bank, as an institution, is the entity that ultimately uncovered the Karnali Development Bank crisis, imposed corrective measures, and referred individuals — including, remarkably, its own former inspection officer — for prosecution. The system worked, eventually. But "eventually" is the word that matters to a shareholder. The Canon Score framework is not a substitute for regulation; it is meant to help an individual investor act earlier than the regulatory process necessarily can, because regulators must build a legal case before they act, while an investor only needs to build a risk judgment.

Lesson 89.3 — The Numbers That Lied: NPL 7.27 Percent Versus 40.85 Percent

This is the centre of the case, and the number is worth sitting with. When Nepal Rastra Bank placed Karnali Development Bank under Prompt Corrective Action in late November 2024, and then formally declared it a troubled, or "problematic," institution in late December 2024, the bank's own reported non-performing loan ratio — the figure it had been disclosing to shareholders and to the market — was 7.27 percent. NRB's own supervisory assessment, once regulators looked underneath the reported numbers, found the true figure to be 40.85 percent.

Read that gap again. Not a rounding difference. Not a modest understatement that a conservative analyst might apply a haircut to. A difference of more than five and a half times. Four out of every ten rupees the bank had lent out were not being repaid on schedule, while the number in the disclosure documents said fewer than one in ten.

Chapters 54 through 58 built the financial-risk side of the Canon Score specifically to catch this kind of thing — not the fraud itself, which by definition is hidden, but the pattern of numbers around a fraud, which is very hard to hide completely. A few of those patterns were visible in Karnali Development Bank's public record before the takeover, for an investor willing to look for them rather than at them.

Consider capital adequacy first. A capital adequacy ratio measures how much of a bank's own capital stands behind its loans and other risk-weighted assets — think of it as the cushion between "a borrower stops paying" and "the bank itself cannot cover its obligations to depositors." NRB requires Class B development banks to hold a minimum ratio, and Karnali Development Bank had already failed to maintain that minimum before the PCA order — this was one of the stated reasons for the corrective action, not a discovery made after it. A bank that cannot maintain its regulatory capital cushion while also reporting a modest, single-digit NPL ratio is presenting two numbers that do not sit comfortably together: if only 7 percent of loans were bad, the capital math should have been much easier to satisfy. That mismatch — thin capital next to a suspiciously clean NPL number — is exactly the kind of cross-check the Canon Score risk framework asks investors to run, comparing a bank's stress indicators against each other rather than reading any single ratio in isolation.

Second, consider the deposit-to-loan relationship. Karnali Development Bank was mobilizing Rs 5.21 billion in deposits against Rs 3.81 billion in loans, and by the time of the takeover, it could not reliably meet deposit repayment obligations — meaning depositors coming to withdraw money were, in effect, running into a liquidity wall. A bank sitting on more deposits than loans should, in ordinary circumstances, have ample liquidity; that combination of "excess" deposits over loans on paper alongside real-world payment difficulty is a strong signal that some of the assets behind those deposits were not what they were reported to be.

Third, and this is the forensic detail that eventually broke the case open publicly: Nepal's Central Investigation Bureau found evidence of financial irregularities exceeding Rs 3 billion, and investigators specifically pointed to a mismatch between the bank's actual deposit records and the figures recorded in its own core banking system — the software platform that is supposed to be the single source of truth for every account balance in the institution. When a bank's core system and its real-world cash position disagree, that is not an accounting judgment call. That is either a severe systems failure or a deliberate falsification, and regulators treated it as the latter.

Lesson 89.4 — Red Flag Detection in Real Time

Chapter 28 built a red-flag checklist meant to be run quickly, almost mechanically, on any stock before it enters a portfolio — a set of questions designed to catch trouble even when an investor cannot independently audit a company's books. Applied retrospectively and in real time to Karnali Development Bank, several of those flags would have triggered well before the December 2024 takeover, for an investor doing the reading rather than relying on a relative's tip.

The first flag is leadership churn without clear explanation. A bank whose chairmanship changes hands more than once within a fairly short span, especially when the departures are not accompanied by a clean, well-explained transition, deserves a closer look. Karnali Development Bank's founding chairman later became a fugitive; a subsequent chairman was a politically connected figure later arrested in the same investigation. Two chairman-level departures under a cloud, within one institution's history, is not a minor governance footnote — it is the single loudest bell the red-flag checklist rings.

The second flag is a bank whose growth story outruns its visible infrastructure. A Class B development bank centred on one of Nepal's less commercially developed regions, expanding its deposit base to over five billion rupees, is a bank whose growth an outside investor should want explained: where is the new lending going, is it diversified across the local economy or concentrated in a handful of large borrowers connected to management, and does loan growth track the kind of collateral and cash-flow discipline NRB requires? Rapid deposit or asset growth at a small, thinly covered institution is not proof of trouble by itself — some regional banks do grow honestly and quickly as they capture underserved markets — but it is exactly the kind of growth that the Canon Score treats as needing a related-party lending check, because concentrated, connected lending is one of the most common mechanisms by which bank capital quietly disappears.

The third flag is disclosure that is technically compliant but practically uninformative. Nepali BFIs are required to publish quarterly financial disclosures and annual reports, and Karnali Development Bank did file these. But filing a report is not the same as filing a report that lets an outside reader actually judge asset quality. A lump-sum NPL percentage without any breakdown by loan category, sector concentration, or borrower size gives an investor almost nothing to independently sanity-check. The Canon Score treats "compliant but shallow" disclosure as a mid-level red flag on its own, and a serious one when it appears alongside the other flags already discussed.

The fourth flag, and perhaps the hardest one for an ordinary retail investor to act on, is regulatory environment risk — the fact that even the supervisory relationship meant to catch problems early can itself be compromised, as the later allegations against an NRB inspection officer suggest happened here. This is why the Canon Score framework never asks an investor to rely solely on "the regulator hasn't said anything" as a green light. It asks investors to treat regulatory silence as neutral information at best, and to keep running their own checklist regardless of whether NRB has issued a public warning yet.

That last sentence is not theoretical. When NRB declared Karnali Development Bank a troubled institution, NEPSE suspended trading in KRBL shares. Anyone still holding the stock at that point was not merely facing a loss — they were facing a position they could not exit at any price, for an unknown length of time, while the bank's true condition was investigated and its assets and liabilities sorted out under NRB-appointed management.

Lesson 89.5 — What "Rescue" Actually Means for Shareholders

It is worth being precise here, because the word "rescue" is doing very different work depending on who you are. When Nepal Rastra Bank takes over a troubled institution under Section 86 of its founding Act, the primary goal, stated explicitly in NRB's own mandate to the appointed management team, is protecting depositors — prioritizing deposit repayment, recovering outstanding loans, and investigating financial irregularities for possible prosecution. Karnali Development Bank's three-member management team, led by a deputy director from NRB's own Bank and Financial Institution Regulation Department, was given exactly this mandate: exercise the powers of the board and the shareholders' general assembly, in that order of priority — deposits first.

Nowhere in that mandate does "protect shareholder value" appear as a stated goal, and this is not an oversight. It reflects the basic hierarchy of claims in a bank failure. Depositors lent the bank their savings expecting it back on demand or on term; shareholders bought an ownership stake expecting to share in profits and to bear the corresponding risk of loss. When a bank's assets turn out to be worth much less than its liabilities — which is exactly what a real 40.85 percent non-performing loan ratio implies for a bank that had reported 7.27 percent — the loss has to land somewhere, and the legal and regulatory framework in Nepal, as in most countries, puts shareholders ahead of depositors in absorbing that loss. A "rescue" of a bank protects the people whose money the bank was holding in trust. It does not promise to protect the people who owned equity in the enterprise that mishandled it.

This is also the moment to place Karnali Development Bank inside the broader pattern of NRB's supervisory activity, because a single dramatic case can leave the impression that this was an isolated event rather than a normal, if serious, part of how Nepal's financial sector is policed. In the months around the Karnali Development Bank takeover, NRB's third-quarter inspection cycle for fiscal year 2081/82 produced a wave of corrective actions across multiple institutions, at varying levels of severity.

The lesson from this table is not that Nepal's smaller BFIs are uniquely dangerous as a class — most Class B and Class C institutions operate for years without ever appearing on a list like this. The lesson is that PCA and its escalations are a regularly used, systemic tool, not an exotic emergency measure reserved for one unlucky bank. An investor holding shares in any thinly covered development bank or finance company should treat "this could happen to my holding" as a live, recurring possibility to underwrite against, not a black-swan tail risk to dismiss.

For an investor thinking about eventual resolution rather than just the crisis moment, it is worth noting what typically happens next in Nepal once NRB has stabilised a troubled institution's operations: recovery of outstanding loans, cleanup of the balance sheet under direct supervision, and — in many though not all cases over the years — an eventual forced or negotiated merger into a stronger institution, sometimes at terms that leave little or nothing for the original shareholders, sometimes with a small residual value depending on how much of the capital shortfall the loan recoveries manage to close. As of the most recent public reporting available at the time of writing, Karnali Development Bank remains under NRB-appointed management, with loan recovery and investigation still underway and no acquiring institution yet named — a reminder that these processes typically take years, not months, to resolve, and that shareholders caught inside a suspended stock have no choice but to wait for that resolution regardless of how it eventually lands.

Lesson 89.6 — Running Karnali Through the Real Canon Score

Now put the actual framework together — Chapter 64's seven dimensions, adjusted for a Class B bank the way Chapter 65's banking guidance directs, using CAR and NPL in place of the generic leverage and profitability checks, related-party lending in place of a generic governance sub-check, and loan-and-deposit growth in place of a generic revenue-growth check — and score Karnali Development Bank honestly, on only the information a disciplined investor could have had before Nepal Rastra Bank acted. That means using the bank's own reported Q1 fiscal-year-2024/25 numbers (the Rs 19.8 million net loss, the 7.27 percent reported NPL ratio, deposits of Rs 5.21 billion against loans of Rs 3.81 billion) and the governance history that was already on the public record by mid-2024 (two chairman-level transitions, one of them combining the chairman and CEO roles, the second a sitting politician). It deliberately does not use the 40.85 percent true NPL figure, which only became public at the moment of NRB's own December 2024 declaration, and it does not use anything from the 2025 criminal investigation — because using those would be scoring hindsight, not the framework.

Financial Strength & Profitability (20 points), adjusted per Chapter 65 to use CAR and NPL rather than generic ratios. Return on equity: a net loss in the most recently disclosed quarter puts this in Chapter 64's bottom band regardless of the loss's small size — 1 out of 8. Capital adequacy, replacing the generic leverage sub-component for a BFI: by the time Q1 FY2024/25 was disclosed, the bank was already failing to hold NRB's minimum capital cushion, the explicit, stated trigger for the Prompt Corrective Action that followed within months — Chapter 64's bottom CAR band — 0 out of 7. Earnings quality and consistency: a disclosed loss sitting next to a suspiciously low reported NPL ratio and a capital shortfall is an internally contradictory picture on its face, before any investigator confirmed why — 1 out of 5. Financial Strength & Profitability: 1 + 0 + 1 = 2 out of 20.

Governance & Promoter Behaviour (15 points), adjusted per Chapter 65 to weight related-party lending and loan concentration heavily for a bank. Promoter and leadership stability: two chairman-level transitions in the bank's history, one combining the chairman and CEO roles in a single person — a real, contemporaneously visible red flag, not a hindsight one — 1 out of 6. Related-party lending and loan concentration: the bank's disclosure gave no breakdown of borrower concentration or related-party exposure at all, a genuine transparency failure in a dimension Chapter 65 says needs the heaviest scrutiny at a bank — 1 out of 5. Disclosure timeliness and quality: compliant but shallow, a lump-sum NPL figure with no sector or borrower-size breakdown, exactly the "compliant but practically uninformative" pattern Lesson 89.4 names — 1 out of 4. Governance & Promoter Behaviour: 1 + 1 + 1 = 3 out of 15.

Liquidity & Tradability (10 points). Karnali Development Bank was, by this book's own account in Lesson 89.1, a stock with "thin to none" analyst coverage and no meaningful trading interest from outside its home region — no verified average daily traded value or free-float percentage could be established from public sources, a genuine data gap for an obscure Class B name rather than a judgment either way. Scored conservatively for both sub-components: 2 out of 5 and 2 out of 5. Liquidity & Tradability: 4 out of 10.

Valuation Reasonableness (15 points). No verified price-to-earnings or price-to-book figure for Karnali Development Bank could be established from public sources at the relevant time — consistent, again, with a stock nobody was covering closely enough to price confidently either way. Scored conservatively: 2 out of 8 and 2 out of 7. Valuation Reasonableness: 4 out of 15.

Sector & Business Model Durability (15 points). A banking license is a real regulatory barrier to entry, but a small, single-region Class B development bank carries none of the scale or diversification advantages of a Class A commercial bank — a moderate moat at best — 4 out of 8. Concentration and dependency: the bank's business was geographically concentrated in one of Nepal's least commercially developed regions, a real structural concentration risk distinct from any fraud — 2 out of 7. Sector & Business Model Durability: 4 + 2 = 6 out of 15.

Growth Trajectory (15 points), adjusted per Chapter 65 to read loan-and-deposit growth rather than generic revenue growth. Loan and deposit growth: deposits of Rs 5.21 billion sitting well above a loan book of Rs 3.81 billion is not, on its own, a sign of healthy growth — Lesson 89.4 already named this exact mismatch as the second red flag a disciplined investor should have checked, because deposit growth outrunning verifiable, well-distributed lending is precisely the pattern that invites a related-party-lending check rather than admiration — 2 out of 8. Earnings consistency: a disclosed net loss in the most recent quarter, with no confirmed multi-year trend either way — 1 out of 7. Growth Trajectory: 2 + 1 = 3 out of 15.

Dividend & Capital Return Discipline (10 points), adjusted per Chapter 65 to weigh a bank's payout against its capital adequacy position. Karnali Development Bank had built exactly the reputation Lesson 89.1 describes — a small development bank retail investors associated with a "long run" of dividends — but a bank already short of its regulatory capital cushion has no business distributing capital to shareholders at all, and no reliable, verified record of the bank halting payouts once its capital position weakened could be established from public sources. Scored conservatively for both the data gap and the CAR-constraint concern Chapter 65 flags specifically for banks: 1 out of 6 and 1 out of 4. Dividend & Capital Return Discipline: 2 out of 10.

Summed, Karnali Development Bank's real Canon Score comes to 24 out of 100 — deep in Chapter 64's Weak/Avoid band on the raw arithmetic alone, and independently capped there by the governance override, since Governance & Promoter Behaviour scored 3 out of 15, well below the 5-point floor. Both paths to the same conclusion, arrived at using only information that was genuinely public before Nepal Rastra Bank acted, is about as clean a "the framework works" result as this book's case studies produce.

The honest answer to this chapter's question, then, is yes, with an important qualification about which parts of the picture were actually knowable in time. A disciplined Canon Score process did not need to predict a criminal case, name a future fugitive, or foresee an NRB inspection officer's 2025 arrest to reach the right conclusion — all of that came later, and crediting an early-2024 decision with foreknowledge of 2025 arrests would be exactly the kind of hindsight error this book warns against elsewhere. What the framework needed, and had, was the governance record — unexplained chairman-level turnover, opaque related-party and concentration disclosure — sitting on the public record for years before the takeover, plus the Q1 FY2024/25 numbers that were already internally contradictory the moment they were filed. The governance dimension alone, scored honestly in early-to-mid 2024, was already low enough to trigger the override and land Karnali Development Bank in Weak/Avoid, well before Prompt Corrective Action in November 2024 and the takeover in December. The specific, damning 40.85 percent NPL figure, by contrast, only became public at the moment of the takeover itself — it confirmed the framework's earlier verdict, but arrived too late to be the thing that triggered an exit, since trading was suspended at essentially the same moment. This is precisely why Chapter 28's red-flag logic treats governance and disclosure failures as reasons to act on detection rather than reasons to wait for financial confirmation: the confirmation, when it comes in the form of a regulatory takeover, typically arrives at the same moment the exit door closes.

Bikash Rai, for what it is worth, never bought the stock. He ran the five questions from this chapter's practical checklist against what little public information existed on Karnali Development Bank in early 2024, found two clear "unclear" answers on the chairman-history question and the related-party disclosure question, and put his notebook away. He was not smart enough, or informed enough, to know what was coming. He simply followed the process the framework asks every investor to follow, and the process did the rest.