Chea Serey, governor of the National Bank of Cambodia, the central bank that licenses and supervises the country’s lenders, gave a new figure for the country’s restructured loans on 27 July 2026. Four other readings of a restructured loan stock sit on the public record, and the five documents carrying them do not count the same thing.
Loan restructuring rewrites a loan’s terms so that a borrower who cannot pay on schedule can. The borrowers holding those accounts carry what the wind-down of the measures produces.
The central bank issued two guidelines in 2025, on 30 July and 10 December, which its Annual Supervision Report 2025, the yearly account of what it found inspecting lenders, records. Under them, fees, penalties, interest and principal repayment were waived. The bank’s Financial Stability Review 2025, its yearly assessment of risks across the financial system, describes the group as displaced civilians, returning migrant workers and frontline soldiers engaged in combat operations, and says the measures run “until March 31, 2026, for refugees and until May 31, 2026, for the soldiers.” Both dates have passed. What falls due for anyone in that group is not stated in any document read for this piece.
A separate measure sits behind the older readings. The Financial Stability Review 2024 records a circular of August 2024 that let lenders restructure a loan twice without changing its classification, the grade a loan carries from performing down to loss, and without setting aside extra provisions against expected losses. That is regulatory forbearance, a rule letting banks delay marking a troubled loan as troubled. The circular ran to the end of December 2025.
The five readings differ on what they count and on whom. Three count accounts, one counts customers, and one gives only a value against total loans. Two cover deposit-taking institutions, lenders allowed to hold the public’s deposits; one covers the whole banking system; one names no reporting population.
FIVE READINGS OF THE RESTRUCTURED STOCK
| Source | Reading attached to | Lenders counted | Unit counted | Figure | Date and population stated |
|---|---|---|---|---|---|
| NBC Financial Stability Review 2024 | 27 Dec 2024 | deposit-taking institutions | accounts | 84,434 accounts, US$1.8 billion | yes |
| International Monetary Fund 2025 Article IV staff report, the fund’s annual assessment of a member economy | mid-Aug 2025 | not stated | value against total loans | US$4.4 billion, about 7.3 percent of total loans | date yes, population no |
| NBC Annual Report 2025 | actual data as of Nov 2025 | whole banking system | customers | KHR 21.1 trillion in riel, Cambodia’s currency, or US$5.3 billion, about 8.3 percent, 252,868 customers | yes |
| NBC Annual Supervision Report 2025 | 2025 report, no as-of date given | deposit-taking institutions, non-deposit-taking reported separately | accounts | 207,830 accounts, US$5.65 billion, 9 percent of total loans | date no, population yes |
| Chea Serey, opening address of 27 Jul 2026, published by the National Bank of Cambodia | not stated | attributed to the banking system, no reporting population defined | accounts | 239,546 accounts, US$4.8 billion, 7.5 percent of total credit | no |
The desk assembled this table from the five documents named in it, and does not present it as every published reading. The World Bank’s Cambodia Economic Update of June 2025 records that around ten percent of total loans were restructured in January 2025 and gives no count basis, which is why it is not in the table.
The National Bank of Cambodia lists four items on its website under 27 July, among them the governor’s opening address and its report on the first half of 2026, both in Khmer. Read on 29 July 2026, the address gives the three figures and states no date they were measured on and no list of lenders inside them. The report’s account of the banking system puts the gross ratio of non-performing loans, those whose borrowers have stopped paying on schedule, at 9.6 percent, and the net ratio, after provisions, at 3.3 percent. It carries no total for the restructured book. Neither document appears on the bank’s English-language news page for that day.
A statement from the National Bank of Cambodia giving the as-of date, the lenders covered and the counting unit for the July figures would make them comparable with the four earlier readings. The bank holds those definitions.
One figure inside the restructured book is already published under the central bank’s own name. Its Annual Supervision Report 2025 records US$1.91 billion of the restructured loans at deposit-taking institutions as non-performing, which the report puts at 33.8 percent of the restructured total at those lenders. That share is measured against the restructured book. The 9.6 percent is measured against all loans in the banking system. The two rest on different bases. The supervision report gives its restructuring figures in a section that opens with the two 2025 guidelines and does not say whether the totals cover only loans restructured under them.
The address says the measures are now being normalised, and gives that as strengthening the resilience of the banking system. Agence Kampuchea Presse, the Cambodian government’s news agency, reported the passage in English the next day as the measures now being gradually phased out to further strengthen resilience. Neither wording names which instrument it reaches. Three windows are dated in the documents read and all three closed before the governor spoke: the August 2024 circular at the end of December 2025, and the relief under the two 2025 guidelines on the dates the Financial Stability Review 2025 gives. A permission window closing is not the same event as a stock of already-restructured loans unwinding, and no document read says which the wording reaches.
The governor gave the change an outcome, that it strengthens the resilience of the banking system. The documents read establish more than one condition that outcome depends on. The table below is the desk’s enumeration from those documents; each entry is what the named document records, not a score.
CONDITIONS THE OUTCOME DEPENDS ON, ENUMERATED BY THE DESK
| Condition | What the named document records |
|---|---|
| Restructured loans are graded at their true classification as the measure lapses | The International Monetary Fund’s 2025 Article IV staff report calls phasing out blanket forbearance by end-2025 critical to timely recognition and resolution of non-performing loans. No document read gives the grading of the restructured stock after the measure lapsed. |
| Capital absorbs the losses that grading recognises | The National Bank of Cambodia’s first-half 2026 report puts the total capital ratio at deposit-taking institutions at 23.3 percent, against a regulatory minimum of 15 percent. No document read measures the loss that grading the restructured stock would produce. |
| Borrowers resume payment | The Annual Supervision Report 2025 records part of the restructured book at deposit-taking institutions as non-performing. It gives no payment status for the remainder. |
| No further shock reopens the need for restructuring | The Annual Supervision Report 2025 records two further guidelines issued in 2025, after the August 2024 circular the Financial Stability Review 2024 records. |
| Credit growth normalises, the National Bank of Cambodia’s own term | The Article IV staff report records the bank telling fund staff that continued use of the forbearance masked vulnerabilities in the banking sector and may have limited the pace of credit normalisation. Customer loans rose 4.6 percent in the first half of 2026 on the bank’s own report, against credit growth of 3 percent in 2024 in the Financial Stability Review 2024. No document read states what rate counts as normal. |
The documents settle one of these conditions and leave the rest open. The report establishes that part of the restructured book remained classified as non-performing; it does not establish whether borrowers in that portion had resumed partial payment or satisfied the conditions for return to performing status. No document read gives the grading of the restructured stock after the measure lapsed, or a measure of the loss that grading would produce, or a rate of credit growth that would count as normal. The change acting alone would establish that the measure has ended, and it would not establish that the loans behind it are performing.