Bank of Bhutan and BNB: Two Ways to Fail

In April 2026 the Royal Monetary Authority fined Bank of Bhutan more than Nu 200 million over governance and control failures surrounding roughly Nu 1.5 billion of erroneous transactions during a core systems migration.

Bank of Bhutan’s statutory auditors, reviewing the same year, issued a clean, unmodified opinion. They recorded no key audit matters and stated expressly that they had not identified areas of specified risk.

Four months earlier, Bhutan National Bank had its FY2025 accounts qualified by its auditors. By the middle of 2026 its capital adequacy ratio had fallen to 11.29%, below the 12.5% minimum the RMA requires, and the regulator had cut its dividend to force it to keep more of its earnings.

Two institutions. Two kinds of trouble. And in each case, one of the two systems designed to catch trouble said nothing was wrong.

That is the subject here — not which bank is worse, but what the pairing reveals about how Bhutanese bank oversight actually works.

The distinction that matters

An external audit and a supervisory inspection ask different questions.

The auditor asks whether the financial statements fairly present the bank’s position under the applicable accounting standards. That is a question about numbers, answered once a year by a firm the bank engages.

The supervisor asks whether the bank is being run properly — whether its controls function, whether its capital is adequate, whether its conduct toward customers is acceptable. That is a question about behaviour, answered continuously by a regulator the bank cannot choose.

A bank can pass one and fail the other. In 2026, Bhutan got a demonstration of both directions at once.

The first way: clean books, failed controls

Bank of Bhutan is the clearer case, because nothing about the accounting was wrong.

Roughly Nu 1.5 billion of transactions were processed erroneously during a core banking migration. In accounting terms the position was reconcilable — the annual accounts were signed without qualification. What failed was the machinery around the numbers: change management, reconciliation, the controls that are supposed to make a migration boring.

The RMA’s response — a penalty exceeding Nu 200 million — is not a statement about solvency. It is a statement about operational risk, and it landed on an institution whose capital position is comfortable. BoB’s capital adequacy ratio stood at 13.85% at the end of 2025 against the 12.5% floor, and rose over the following half-year.

So: a bank in no financial danger, with clean audited accounts, penalised heavily for how it operates. The auditors were not wrong. They were answering a different question.

The RMA has since opened a review of Bhutan National Bank over double-deducted charges affecting more than 54,000 accounts — again a conduct-and-controls matter rather than a solvency one, and again the kind of thing an audit opinion is not designed to surface.

The second way: clean conduct, broken capital

BNB runs the opposite way, and it is the more serious of the two.

Here the auditors did not sign cleanly. The FY2025 statutory audit was qualified — the formal signal that the auditors could not satisfy themselves on something material.

By the half-year disclosure to 30 June 2026, the numbers had moved sharply:

The regulatory minimum is 12.5%, plus operational-risk capital. BNB’s headline capital ratio is now below it.

The bank’s own disclosures carry an explicit line for the cause — “Losses for the Current Year” — of Nu 1,665.8 million.

Two explanations that do not survive contact with the filings

When a bank’s capital falls this fast, the reassuring explanation is usually that nothing real has happened — that an accounting change moved the numbers rather than the business. Two versions have circulated, and the filings dispose of both.

First: that adopting Bhutanese Accounting Standards created a one-off transition hit. BNB’s own quarterly disclosure shows the opposite. The transition raised retained earnings — by Nu 645.7 million on the BAS basis against the prior comparative. A transition adjustment that increases equity cannot be the source of a capital hole.

Second: that expected-credit-loss accounting is itself driving the losses. This one is disposed of by Bank of Bhutan. BoB has reported on the same basis for years; its half-year statements in 2025 and 2026 carry the identical accounting footnote; it booked substantial credit impairment in both FY2024 and FY2025; and its 2025 comparatives are unchanged between the as-published statement and the following year’s document. Zero restatement is the signature a genuine transition does not leave. What actually changed on 1 January 2026 was the RMA’s disclosure template. Both banks switched it on the same date. One stayed profitable.

The accounting is not what happened. Something in BNB’s loan book is.

What the regulator did about it

The most telling detail sits in BNB’s own annual report, in the section on dividends.

The board recommended the maximum permissible dividend — Nu 1.458 per share, or 14.58% — as allowed under the RMA’s Prudential Regulations. The RMA did not approve it. It approved a payout ratio of 35%, equivalent to Nu 0.927 per share, or 9.27%.

Read plainly: the bank proposed distributing the maximum its regulator’s own rules allowed, and the regulator told it to retain roughly a third more of its earnings than it had asked to keep. That is a supervisor using the one lever that directly rebuilds capital, and it is the clearest signal in the public record of how the RMA reads BNB’s position.

The eleven-year lesson underneath

Both cases sit on top of a slower problem, and one recovery illustrates it.

A resort property in Trongsa, carried at around Nu 280 million, stood against roughly Nu 179.5 million of Bank of Bhutan exposure. It was eventually transferred for Nu 9.5 million — about five percent of the principal — after some eleven years of legal process.

That is the real cost of impaired lending in a small jurisdiction: not the write-down, but the decade. Collateral that cannot be realised at anything like its carrying value, on a timescale no provisioning model assumes.

It matters now because the loan books are turning. BoB’s housing non-performing loans rose from Nu 389.6 million to Nu 686.3 million across 2025 — up 76% on a housing book that grew just 3.6% — while its total NPLs rose 61.5%, from Nu 1.75 billion to Nu 2.82 billion. Those are the audited figures, from the bank whose audit was clean. Eleven years is the context in which to read them.

What this is not

It is worth being precise about the limits of what the public record supports.

Neither bank is failing. Bank of Bhutan has ample capital and a conduct problem. Bhutan National Bank has a capital problem and, on the record, no equivalent conduct finding — the RMA review of its charging is open, not concluded. Bhutan’s banking system as a whole reported record profits in FY2025, and profits look backward while asset quality looks forward, which is exactly why the two can diverge for a period without either being wrong.

Nor is any of this hidden. Every figure above comes from documents the banks and the regulator publish precisely so that it can be read. The disclosure regime is working. What the two cases show is that the reader has to hold two records at once, because neither system sees the other’s failure mode.

A clean audit is not a clean bill of health. It is an opinion about the accounts, from a firm engaged to give one. And a strong capital ratio says nothing about whether a bank can execute a systems migration without losing track of Nu 1.5 billion.

In 2026 Bhutan learned both halves of that lesson, four months apart, from its two largest banks.