A banking system does not lose half its capital in a profitable year
In 2024 Bangladesh's banks reported a profit, and reported capital adequacy falling from 13.15% to 5.59%. Both cannot describe the same year. The arithmetic says what happened: the losses were already there, and the accounts had not said so.
Bangladesh’s banking system reported two things about 2024 that cannot both describe the same year. It reported a return on assets of 1.19% and a return on equity of 7.36%: a profitable year, thinner than 2023’s 1.68% but comfortably in the black. And it reported that regulatory capital fell from 13.15% of risk-weighted assets to 5.59%, with Tier 1 capital dropping from 10.29% to 3.25% and common equity Tier 1 from 9.70% to 2.72%.
A profitable year does not destroy half of a capital base. Retained earnings add to capital; they do not subtract from it. So either the profit is not what it appears, or the capital was never what it appeared. The rest of the numbers say which.
| Year | Capital / RWA | Tier 1 | NPLs | Provision cover | Uncovered NPLs / capital | ROA |
|---|---|---|---|---|---|---|
| 2019 | 11.57 | 7.69 | 8.90 | 42.70 | 44.62 | 1.16 |
| 2020 | 11.64 | 7.42 | 7.74 | 48.76 | 34.49 | 0.83 |
| 2021 | 11.08 | 7.44 | 7.97 | 47.15 | 39.42 | 0.74 |
| 2022 | 13.32 | 10.26 | 8.72 | 44.98 | 38.76 | 1.44 |
| 2023 | 13.15 | 10.29 | 9.57 | 41.29 | 47.44 | 1.68 |
| 2024 | 5.59 | 3.25 | 18.96 | 25.63 | 271.97 | 1.19 |
All figures percent, IMF Financial Soundness Indicators, as published. No FinObservatory adjustment.
The number that carries the story
Non-performing loans doubled, from 9.57% of gross loans to 18.96%. That alone would be a bad year. What makes it something else is that provision coverage fell at the same time, from 41.29% of NPLs to 25.63%: the system recognised twice as many bad loans and set aside a smaller share against them.
Put those together and you get the figure that matters, the one the headline ratios bury. Non-performing loans net of provisions, measured against capital, went from 47.44% to 271.97%. The bad loans the system has admitted to, and not yet provided for, are now 2.7 times its entire capital base. In 2023 they were about half of it.
That is the whole of the arithmetic. If those loans were written down to what they are worth, the capital would not be thin. It would be gone.
It was not a liquidity event, and it was not a bad year of lending
Two things did not happen in 2024, and both are worth stating because they narrow the explanation to one.
Liquidity did not deteriorate. The liquidity coverage ratio rose, from 147.69 to 157.50, and liquid assets as a share of total assets rose too. Whatever this was, the banks were not short of cash, and depositors were not being turned away at the counter. A capital number can collapse while the tills stay full, and here it did.
And the system did not trade its way into the hole. It earned 1.19% on assets. A bank that loses half its capital in a year reports a deep loss, not a profit. The 2024 profit is itself part of the evidence: with provision coverage falling to 25.63%, less was charged against earnings than the newly recognised bad loans would warrant, which flatters the profit in exactly the year the capital broke.
What is left is the only explanation that fits every number at once: the loans were already bad, and the accounts had not been saying so. The step-change is in the reporting, not in the lending. 2024 is the year the numbers started telling the truth, which means the years above it in that table were not.
How far outside normal
Among the 152 countries reporting both capital adequacy and non-performing loans to the IMF, Bangladesh now has the third weakest capital position in the world and the fourth highest bad-loan ratio.
Only 2 countries report weaker capital: Equatorial Guinea (-18.87%, 2023), Chad (-1.39%, 2023). Both are reporting a banking system whose capital is already negative. Bangladesh is not in a peer group with its neighbours on this measure. It is in a peer group with systems that have already failed.
One honest caveat on that ranking: the IMF’s countries do not all report to the same date, so this compares the most recent picture each country has published, not a synchronised cross-section. The stalest entry in the set is Vanuatu, whose last reported year is 2017. It is a ranking of what is on the record, which is the only ranking there is.
What this brief does not say
It names no bank. FinObservatory’s Bangladesh layer is system and category aggregates only, by standing policy, and nothing here should be read as a statement about any individual institution: the distress is certainly not spread evenly across a system of this size, but the aggregates cannot tell you where it sits, and neither can I.
It also makes no forecast. Nothing above says a crisis follows. Recognition is what a system does on the way out of a problem as well as on the way into one, and a capital ratio that finally reflects reality is better than one that does not, however much worse it looks. What the numbers establish is narrower and harder: the bad loans in Bangladesh’s banks in 2024 were not created in 2024, and the ratios published before then did not describe the system that existed.
Every figure above is IMF Financial Soundness Indicators, queried from the FinObservatory estate when this page is built and reproduced as the IMF publishes it. None of it is typed into the prose. The series is on the Bangladesh page; the source and its licence are in the data catalog.