
Bangladesh Bank has published a five-year, 13-point roadmap to clean up the country's bad loans, presented to parliament's finance committee. The plan caps single-borrower exposure across the banking system, moves to expected-credit-loss provisioning, names wilful defaulters in public, and pushes large financings into the bond market. It is the most detailed bad-loan plan any Bangladesh central bank has put on paper. Its own numbers explain why it may not work: Tk2,22,357 crore in required provisions were never set aside.
What the report admits
The stock first. Non-performing loans stood at Tk6,06,555 crore at end-June, 32.78 percent of outstanding loans. A third of all credit disbursed in Bangladesh is not performing. Against that stock, the provisioning arithmetic: Tk4,83,397 crore required, Tk2,61,039 crore set aside. The shortfall, Tk2,22,357 crore, is larger than most annual development budgets and it is the number the rest of this article is about.
| Indicator | Figure | As of |
|---|---|---|
| NPL stock | Tk6,06,555 crore (32.78% of outstanding loans) | June 2026 |
| Provisioning required | Tk4,83,397 crore | June 2026 |
| Provisioning set aside | Tk2,61,039 crore | June 2026 |
| Provisioning shortfall | Tk2,22,357 crore | June 2026 |
| Banks with accumulated losses | 22 banks, Tk2,53,938 crore | March 2026 |
| Sector capital adequacy ratio | Negative 3.17% (minimum 12.5%) | March 2026 |
| Banks meeting CAR minimum | 40 | March 2026 |
| Shariah-based banks CAR | Negative 44.77% | March 2026 |
| Lender-of-last-resort liquidity | Tk84,945.71 crore to 12 banks | 27 Sep 2026 |
| Sector operating profit vs loss | Tk789 crore profit against Tk8,417 crore loss | 2026 |
| Return on assets / equity | Negative 1.03% / negative 44.22% | 2026 |
Twelve banks had drawn Tk84,945.71 crore in lender-of-last-resort liquidity as of September 27, interest-only with principal rolled over as demand loans; Islami Bank's Tk17,000 crore was the largest single draw. The sector's capital adequacy ratio stood at negative 3.17 percent in March against the Basel III minimum of 12.5 percent. Only 40 banks met the minimum. Shariah-based banks sat at negative 44.77 percent. These are not the figures of a banking system with a bad-loan problem. They are the figures of a banking system that is, in aggregate, insolvent on a mark-to-market basis and liquid only because the central bank keeps the window open.
Why the hole comes first
The forensic point is sequencing. Every long-term measure in the 13 points assumes capital that does not exist.
Expected-credit-loss provisioning under IFRS is a recognition technology. It forces banks to book losses earlier, which is honest, and which immediately deepens the capital hole it is meant to manage. A single-borrower system-wide exposure cap is a prevention technology. It stops the next Tk17,000 crore concentration but does nothing about the Tk6.06 lakh crore already on the books. Publishing wilful-defaulter lists is a shaming technology. It works only if the legal system can then seize and sell, which the roadmap's own writ-petition curbs admit is currently blocked. Private asset management companies are a disposal technology. They buy bad loans at a discount, which crystallizes the loss the provisioning hole is currently hiding.
Each of the 13 points is aimed at the flow of future bad loans or the plumbing of recovery. None of them fills the Tk2.22 lakh crore stock hole. A roadmap that sequences process before capital is a timetable for a repair that has not been funded.
Power exposure runs deep
The power sector's loan book, estimated at Tk1.49 trillion in this newsroom's forensic audit of the sector's contractual architecture, is about 24.6 percent of the Tk6.06 lakh crore NPL stock. That is a scale comparison, not an attribution: it measures how large power-sector exposure is next to the bad-loan stock, not how much of the stock sits in power. The sector's own non-performing share is not established by these figures.
The roadmap's single-borrower cap, applied backward, would have caught the independent power producer concentration before it compounded. Applied forward, it is the one long-term measure with teeth for the power book. Everything else in the 13 points treats the power-sector stock as someone else's problem.
Correction, 5 Oct 2026: an earlier version of this article divided the entire Tk1.49 trillion power-sector loan book by the Tk6.06 lakh crore NPL stock and stated that roughly a quarter of the country's bad loans sit in the power sector. That conclusion is not established unless the power loan book is itself all non-performing. The comparison measures exposure scale, not the sector's share of bad loans.
Figure note, 5 Oct 2026: this article uses Bangladesh Bank's June NPL ratio of 32.78 percent. Earlier newsroom coverage cited 32.26 percent, a different reference period.
What would change the read
Stated so they can be proven wrong. The roadmap graduates from wish list to repair if four things print: the provisioning shortfall narrows in quarterly disclosures instead of widening; sector capital adequacy returns to positive territory; a top-20 defaulter faces actual enforcement under the new exposure cap; the first private asset management company completes a portfolio purchase at a disclosed discount. If none of those print within the short-term window of one year, the 13 points were process.
Sources
- The Business Standard, "BB to limit borrowers' sector-wide exposure, target each bank's top defaulters to tame bad loans," 5 Oct 2026.