On 15 September the central bank published the August issue of Monetary conditions and the transmission of monetary policy. It raises two forecasts at once. The forecast for the banking sector's structural liquidity deficit at the end of 2026 rises by 1.6 trn roubles, to 4.0–5.2 trn. Banks' projected borrowing need at the central bank's own auctions in the December averaging period rises by 2 trn, to 8.0–9.2 trn.
The bank gives one reason, and only one:
“The forecast was changed mainly because demand for cash increased, taking into account its actual path since the start of the year.”
The size of that demand
The August comment does not give an annual figure for cash. Its own table 4 does. The line “change in cash in circulation”, measured by its effect on bank liquidity, shows an actual −1 trn roubles for 2025, −2.7 trn for January–August 2026, −0.4 for August alone, and a 2026 forecast range of −4.0 to −3.4 trn.
The point estimate inside that range is in the other document the comment cites, the draft monetary policy guidelines for 2027–2029.pdf) of 31 August, approved by the bank's board on 28 August. Table 2 of its annex gives the same line as −1.0 trn in 2025 and −3.7 trn in the 2026 baseline, then −2.0, −1.6 and −1.6 for the three years after. Table 1 shows the same thing as a level: cash in circulation outside the central bank goes from 19.4 trn roubles at the end of 2025 to 23.1 trn at the end of 2026 — a gap of exactly the 3.7 trn the flow line carries.
So this is not simply a year worse than the one before it. On the central bank's own baseline, cash takes out of the banking system in 2026 3.7 times what it took out in 2025, and the bank has written that into the scenario it plans policy on.
The August move itself is small and points the same way. Cash in circulation rose 0.4 trn roubles over the month. The bank gives the seasonal reason — holidays, the run-up to the school year — and then adds the sentence that matters: “This is higher than in previous years.” The share of cash in the money supply held at 15.1%.
What the same table says about the budget
It is worth being exact about what is not happening here, because the obvious suspicion is the wrong one on this document's numbers.
In the same table 4, the line for the extended government's balances at the central bank — the budget's own effect on bank liquidity — added 1.2 trn roubles in 2025, 0.8 trn over January–August 2026 and 0.1 trn in August, and its 2026 forecast is a range of −0.2 to 0.2, that is, approximately nothing. Over January–August the total drain on liquidity was 2.0 trn, of which cash alone was 2.7 trn; the budget offset part of it.
The budget channel does show up elsewhere in the document, and there it is shrinking. In July the contribution of claims on the economy to the annual growth of broad money rose to 12.0 percentage points from 11.2, while the contribution of budget operations fell to 4.7 points from 5.1. Annual money-supply growth slowed: M2 by 0.2 points to 13.0%, M2X by 0.4 points to 12.6%.
Where the money goes when it leaves
The cash is coming from households. In July the annual growth of household rouble funds slowed to 9.9% from 10.9%. Current-account balances grew while the term segment saw an outflow, and the bank names the destination directly: household funds declined because “some were being shifted into cash”. M0's share of M2 rose to 15.0% from 14.7%. In August, preliminary data showed a slight decline in household rouble account and deposit balances.
A deposit that becomes a banknote does not stop being someone's savings, but it stops being bank funding. And bank funding is what buys government debt. In August Minfin ran no OFZ auctions at all — the pause announced on 20 July held for the whole month — and the largest banks were the main net sellers on the secondary market, at −100 bn roubles, where trading activity roughly halved against July. The ten-year yield ended August at 16.04%, up 35 basis points, with the ten-year-to-two-year spread widening to 159 points from 111.
That pause has since ended. On 24 August Minfin allowed for a return to auctions in September if the market stabilised and rates were acceptable to the budget, and on 2 September it held its first auction after the pause, of floating-rate OFZ. The two floating-rate issues registered in July and available for placement, 29030 and 29031, carry 1.5 trn roubles at par.
What replaces the departed deposits in the meantime is the central bank. Banks' average need for liquidity on market terms rose to 5.2 trn roubles in August from 5.0 trn in July; the forecast for the December averaging period is 8.0–9.2 trn. That is funding at a key rate of 14.00% — no rate meeting was held in August, the rate having been cut 25 basis points on 24 July, and it was held again on 11 September — rather than deposits at the 13.1% short and 10.6% long rates of July.
What the bank says this does not mean
The comment is careful, and the caution is its own:
“The gradual increase in the structural liquidity deficit is an expected trend. Its size relative to the assets of the banking sector remains insignificant, at 1.2%, which is many times smaller than the values observed from the second half of 2012 to the beginning of 2016. The growth of the structural deficit has no effect whatsoever on the stability of the banking sector. The effects of this growth are fully covered by the Bank of Russia's liquidity management operations.”
We have no basis to dispute that, and this piece does not. A liquidity deficit is not a solvency problem, and the central bank can and does close it daily. The question this site asks is a different one: not whether the banks stand, but what the state's borrowing costs and where the money to lend it comes from. On that, the composition matters. Funding that used to be a depositor's balance at 10–13% is increasingly a refinancing operation at 14%, and the institution supplying it is the same one that sets the rate.
What this changes in our model
Nothing directly. An analytical commentary is not one of our eight components, and no forecast enters the index.
It sits closest to money supply (M2), which scores 86 on 12.9% annual growth — the structural point here is that the composition of that aggregate is shifting towards cash, which the growth rate alone does not show. Today's data update moved one indicator: the liquid NWF rose from $46.2 bn to $46.7 bn on August data, taking that score from 28 to 29. The strength index stands at 47.3 out of 100, in the stress zone, against 47.2 yesterday. No trigger.
What we do not know
Why cash demand is rising. The bank gives a seasonal explanation for August and states that the dynamics since the start of the year have run above normal. There is no breakdown by region, denomination or type of recipient anywhere in the document. Attributing this to payments to soldiers or to front-line regions cannot be done from these papers, and we do not do it.
Whether the 3.7 trn figure has been revised since 31 August. It comes from the guidelines draft of that date, which was itself prepared on statistics as of 21 August. The August comment says the actual dynamics turned out higher, and raises the deficit range accordingly, but publishes no new cash number of its own.
How much of the December 8.0–9.2 trn is the budget. Table 4's own government line forecasts approximately nothing for 2026, but that is a forecast about balances at the central bank, not a statement about spending, and December is when budget flows are largest. It is worth checking again against the actual data.
What the August breakdown of broad money looks like. The 12.0 and 4.7 point contributions are July figures. For August the bank publishes only an operational estimate of the aggregates.