On 14 September the central bank published its monthly Monitoring of Sectoral Financial Flows, issue No. 9 (114), covering payments through the Bank of Russia's own payment system to 31 August. Its opening sentence reports growth. Twenty pages later, a table shows incoming payments to state-demand industries 20.0% lower than a year earlier, while payments to consumer-facing industries are 12.5% higher. The bank's summary leads with the first figure; the second sits in table 10.

What the bank published

The first page:

“In August 2026 the volume of incoming payments made through the Bank of Russia payment system increased by 2.6% compared with the average level of the second quarter of 2026. Excluding extraction, petroleum products and public administration, incoming payments grew by 5.6%. Growth in receipts was recorded in the industries of consumer, intermediate and external demand.”

That second figure does not isolate the state, because it also strips out extraction and petroleum products. Table 1 supplies the line that does: excluding extraction and petroleum products alone, the flow grew 2.0%. Adding public administration to the exclusions takes it to 5.6%. Those 3.6 points are the state's weight on the total. Two paragraphs later, in the same summary:

“The volume of receipts in the industries of state demand fell by 8.5% in August. The main contribution to the decline continues to be made by the activity of public administration bodies (84).”

The number inside

Table 1 compares each month with the previous quarter's seasonally adjusted average. Table 10 makes the simpler comparison with the same period a year earlier. Read down its state-demand row: +2.4% in the third quarter of 2025, +1.2% in the fourth, then −2.2% in the first quarter of 2026, −14.0% in the second, −18.7% in June, −22.6% in July and −20.0% in August. The sign changed at the turn of the year, and the gap has widened every quarter since.

For the rest of the economy in the same month, the same table shows consumer demand at +12.5%, intermediate demand at +8.4% and external demand at +4.0%. Investment demand is the other negative, at −12.0%, after −13.7% in July and −14.2% in the second quarter. The aggregate flow, weighted by industry shares in value added, was +3.3% year on year; excluding extraction, petroleum products and public administration, it was +5.5%.

One industry does the work

Table 3 breaks the state-demand group into contributions weighted by each industry's share of the group. Class 84 in the activity classification — “activity of public administration bodies” — carries a weight of 60.4, and its contribution in August was −19.4 percentage points, after −18.5 in July and −11.8 across the second quarter. In the fourth quarter of 2025, the same contribution was +3.9.

Table 9 gives that class's own rate against the previous quarter's average: −32.1% in August, −30.6% in July, −17.5% in June, −26.8% in May.

The school year partly offsets it. Education (85) contributed +10.1 points to the group in August, on its own rate of +55.8%, and healthcare (86) contributed a further +0.5. Both are financed mainly from regional budgets, and both are positive. The fall sits in class 84, which the bank does not split between federal and regional bodies.

What this series is, and what it is not

The bank is explicit about the limits, and they matter more than usual here.

Coverage is partial: these are payments through the Bank of Russia's system, and “they cover less than half of the payments made, taking into account payments on LORO/NOSTRO accounts, intra-bank transfers and transfers through other payment systems”.

The data are winsorised. For each industry a threshold is set at “the 99.5 percentile of the distribution of non-zero payments”, and “payment values exceeding this threshold are replaced by the value of the established boundary”. Large one-off budget transfers therefore do not enter the series at full size. This measures the breadth and frequency of payments more than their total volume, and a reader who takes −20% as “state spending fell by a fifth” has read it wrong.

The weights changed mid-series: “until July 2026 the base year for calculating value-added shares was 2018. After that, a transition was made to calculating weights with 2023 as the base year.” The document does not show how much of the shift in level came from that change alone.

Attribution is by the recipient's main activity code in the state register. The bank does not split class 84 into defence and civil, or federal and regional. Nothing in this document says these are military payments, and we do not say it.

Why a payment series is a war-finance question

Because of what it sees. Minfin's August execution table, published this month, reached us on the same timetable — we read it today — but it reports cash totals by budget function, not money arriving in the accounts that pay contractors. The series that measures the queue directly, Rosstat's overdue receivables, comes a quarter later, and Rosstat has been unreachable for six days. Incoming payments to public administration bodies are money landing in the accounts of institutions that then pay their contractors. When that flow runs a third below the previous quarter's average for two months while the consumer sector grows at 12% a year, the reading is not a general cooling but a queue.

That queue is where our weakest indicator comes from. It is also the same seam we found in the CBR's own banking review, which tied faster corporate lending to slower budget spending, and in the 580.4 bn roubles that national projects spent in a single August. A payment deferred is not a payment saved.

What this changes in our model

Nothing. A payment-flow series is not one of our eight components, and no analytical publication enters the index.

Overdue receivables remain the weakest at 9 out of 100, on data to the end of June. The federal deficit scores 65 at 3.29% of GDP over the rolling twelve months to August — up from 3.27% in today's data update — and regional budgets improved from −1.616 to −1.509 trn roubles, lifting that score from 28 to 31. The strength index stands at 47.2 out of 100, in the stress zone, against 46.8 yesterday. No trigger.

What we do not know

Whether this is a cut or a delay. The series records payments, not obligations. Money not paid in August may be paid in October; a deferral and a cut look identical here. This is precisely why we watch it as a leading signal for receivables rather than as a spending figure.

What fell inside class 84. The bank publishes the class, not its contents. Wages, procurement, transfers to contractors and inter-budget flows sit in the same line.

How much of the drop is the new weights. The base year moved from 2018 to 2023 in July 2026, in the middle of the comparison we are reading. The document does not quantify the effect.

Whether the fall continues. The next issue is monthly, and September's figure will show whether August's −20.0% is a floor or a step.