On 19 November the Bank of Russia published its banking sector analytical review for the first nine months of 2026; the data are cut off on 4 November. We read the 54-page file in this session. Corporate loans showing signs of problems stood at 13.6 trillion roubles on 30 September — 13.2% of the corporate loan book, against 10.6 trillion and 11.2% at the end of 2025. The page carrying those figures is headed “the quality of corporate loans worsened, but they are adequately covered by provisions and collateral”. In the same document, that cover fell from 51.4% to 44.9% over the same nine months.
Both sides of that contrast come from the review. The second is worth following because of where the growth came from.
The formally bad part shrank
The Bank of Russia's category is broader than default. It combines loans in quality categories IV–V — the formally bad ones — with two analytical additions of its own: “risky restructurings” and “other loans with signs of problems”. Each is defined in a footnote listing five numbered conditions. The appendix prints the three parts separately as shares of the corporate book:
| Part of the book | 31.12.25 | 30.09.26 | Cover by provisions and collateral, 30.09.26 |
|---|---|---|---|
| Quality categories IV–V | 4.3% | 4.1% | 85.0% |
| Risky restructurings | 4.3% | 5.6% | 32.8% |
| Other loans with signs of problems | 2.5% | 3.6% | 18.1% |
| All loans with signs of problems | 11.2% | 13.2% | 44.9% |
The parts are rounded to a tenth and do not add exactly to the totals printed beneath them. The share in categories IV–V did not rise. It fell, by 0.2 percentage points, and the review says so in as many words. All the growth was in the two analytical categories — from 6.8% of the book to 9.2% — and those are the thinly covered categories: 32.8% and 18.1%, against 85.0% for the formally bad loans.
That accounts for the entire fall in cover. No category's own coverage ratio collapsed; risky restructurings were covered 35.7% at the end of 2025 and 32.8% now, the other category 20.4% and 18.1%. The mix shifted towards the thinly covered end, pulling the total down 6.5 points. A sentence saying the book is adequately covered and a sentence saying cover fell to 44.9% can both be true of the same page, and the review prints both.
A trillion of it is last year's
The review gives the nine-month movement in two forms. The stock of problem debt “increased by 2.0 trillion roubles” over the period; its share of the book rose “by 1.2 percentage points, taking account of the correction on old loans”. Neither movement reconciles with the published figures at the start and end of the period.
10.6 + 2.0 = 12.6, and the stock on 30 September is 13.6. 11.2 + 1.2 = 12.4, and the share on 30 September is 13.2.
The gap is a trillion roubles and 0.8 of a point, and the review explains it in a footnote to the growth figure: the increase “is given taking account of the restatement of reporting by individual banks that reflected a correction on old loans”. The sectoral table prints the same thing from the other side — it gives the nine-month change as +18.6%, which on a stock of 13.6 trillion implies a starting point near 11.5 trillion, not the 10.6 published for the end of 2025.
Read forwards, this makes the deterioration smaller than the two published stocks suggest: 2.0 trillion over nine months, not 3.0. Read backwards, our arithmetic says the end-2025 figure was about a trillion roubles short of what the same loans look like once the banks restated them. The correction enters only at the current point; the published historical series remains 6.9 trillion for 2022, 8.4 for 2023, 8.9 for 2024 and 10.6 for 2025.
The relief was extended, not withdrawn
The review names the cause as the financial position of particular borrowers: real estate added 0.7 trillion roubles of problem debt over the nine months, manufacturing 0.4 trillion and metals 0.4 trillion. Restructured corporate loans as a whole now stand at 19.0 trillion roubles, 18.5% of the book, up from 16.3% at the end of 2025.
At the same time, the regulator extended its relief. So that banks do more to help borrowers “with the potential to restore stability”, as the review puts it, the Bank of Russia extended its easing of provisioning requirements on restructured loans to such borrowers until 1 January 2027. That is a deferral of recognition, not a write-down and not a loss: the loans stay on the books and the provisions that would otherwise stand against them are not required yet. It also means the next nine-month comparison will be drawn against a book still carrying the easing.
Where this touches the budget
The part covered by neither provisions nor collateral is 7.5 trillion roubles. The Bank of Russia sets that against the sector's capital and calls the risk manageable, in its own formulation: the uncovered part is “up to ~75% of the capital buffer above the regulatory minimums”, the risks “are realised gradually, and some borrowers will be able to settle their debt”. The sector remains profitable: profit for the nine months was 3.6 trillion roubles, 36% more than a year earlier, and the forecast for 2026 was raised to 4.5–4.9 trillion. Regulatory capital stands at 22.6 trillion roubles and the capital adequacy ratio at 13.7%, both higher than a year ago.
What the buffer does is bound how much more risk the banks can carry, and the banks are the domestic buyers of the state's debt. The review does not draw that line, and neither do we: an uncovered 7.5 trillion sitting against the buffer is consistent with a tighter limit on what the sector can absorb next, not proof that it has been reached.
The nearer trace is in the real economy. This is bank reporting on its borrowers, while Russia's overdue corporate receivables count bills that Russian firms have not paid each other, published by Rosstat under a different definition. The company that stops paying its supplier and the company whose bank reschedules its debt are often the same company, but the two series measure it from opposite sides. Neither is a reading of the other: the trace here is second-hand. Two days ago the CBR's own survey of enterprises showed firms reporting tighter credit terms for a third consecutive month, specifically in non-price conditions: the borrowers' account of the same autumn.
What the review does not say
The review does not separate how much of the 2.0 trillion is new deterioration from how much is existing exposure crossing into the analytical categories. It gives the sectoral split of problem debt in its appendix — real estate at 4.9 trillion roubles, oil and gas 2.5, metals and coal 1.8 — but no line for defence-sector borrowers, who are not a category in its industrial classification. The definitions of “risky restructuring” and “other loans with signs of problems” rest on analytical judgements by the regulator's own department, set out in the footnotes; we have not tested them against the underlying reporting. And the data stop on 4 November, so October is incomplete in some of the cuts.