On 9 September Minfin sold fixed-coupon government bonds for the first time since 1 July. Its own release puts the day's placement at 90.660 bn roubles “at nominal”. Cash proceeds were 66.148 bn. The two figures measure the same auctions in different ways, and the borrowing programme uses the first. Across the year to 3 September, Minfin's own summary table shows a 450.0 bn-rouble gap between them.
What Minfin published on 9 September
Three separate notices appeared on Minfin's auction page. Issue 26218RMFS, maturing 17 September 2031, drew 91.824 bn roubles of demand; 51.496 bn was placed for proceeds of 42.152 bn, at a cut-off price of 77.9095% of par and a weighted average yield of 15.45% a year. No additional placement followed.
Issue 26230RMFS, maturing 16 March 2039, drew 72.718 bn of demand:
“Results of the placement of issue No. 26230RMFS: — volume of demand – 72.718 bn roubles; — placed volume of the issue – 35.215 bn roubles; — proceeds from the placement – 21.576 bn roubles; — cut-off price – 57.8152% of par; — yield at the cut-off price – 15.98% a year; — weighted average price – 57.8526% of par; — weighted average yield – 15.97% a year.”
A third notice records the additional placement after that auction: 3.949 bn more of 26230RMFS out of 21.641 bn offered, for proceeds of 2.420 bn, at the auction's weighted average price.
In total, the day drew 164.542 bn of demand, added 90.660 bn of principal to the state's debt and brought in 66.148 bn of cash — 73.0% of nominal.
Two ways of counting
A borrowing programme is counted by the face value of the bonds sold, because that is what has to be repaid. The budget, however, receives what the buyer actually pays. When a bond sells below par, those are different numbers, and on 9 September they were 24.512 bn roubles apart.
The proceeds figure itself includes accrued interest, and Minfin's documents let us prove it. For 26230RMFS, price times nominal comes to 20.373 bn against stated proceeds of 21.576 bn — a difference of 3.42% of nominal. The additional placement of the same issue, on the same day, shows exactly the same 3.42%. That is accrued coupon interest: the buyer advances the coupon that has built up since the last payment date and gets it back at the next one. Strip it out and the day raised 62.783 bn against the principal itself, 69.25% of nominal.
None of this is a loss. A bond sells below par because its coupon is below the yield the market now demands. The state pays that lower coupon every year, then repays the full face value at maturity. The discount is a deferred cost, not money lost — the same distinction we drew about tax deferral in August. But nominal placement and cash proceeds are not interchangeable in a headline. “90.66 bn placed” and “66.15 bn received” describe the same event, and only one of them is money the budget can spend this year.
The year in Minfin's own table
Minfin publishes every auction of the year in one workbook, updated to 3 September. Its “total” line reads: demand 6,807.3 bn roubles, placed at nominal 4,262.8 bn, proceeds 3,812.8 bn. That is 89.4% — and a gap of 450.0 bn roubles between the principal Russia has taken on this year and the cash it has received for it. Adding 9 September brings the year's gap to 474.5 bn.
Why the day looks worse than the year
It would be easy to read 73.0% against 89.4% as evidence that the market turned against long fixed-coupon debt this week. The same table says otherwise. The ratio mostly reflects the issue being tapped, and it has been stable all year. Issue 26230RMFS returned 64.1% of nominal in January, 66.6% in March, 63.0% in May and 61.3% on 9 September. Issue 26218RMFS returned 82.4% in March, 81.4% in April and 81.9% on 9 September. Neither moved much.
What moved was the mix. 26230RMFS — the deep-discount 2039 paper — was 39.164 bn of the day's 90.660 bn, or 43.2%. A day weighted towards that issue produces a low ratio without anything new happening to the price of Russian debt.
Nor is the year's 89.4% the work of one large placement. Excluding the trillion-rouble floater of 2 September, the year runs at 88.1%, almost unchanged. The discount is ordinary, persistent and reflects which bonds Minfin has left to sell.
One trillion in one day
The same table shows how the year's nominal total was assembled. On 2 September, floating-coupon issue 29031RMFS, maturing 29 July 2042, drew 1,424.700 bn of demand. Minfin placed exactly 1,000.000 bn at a cut-off price of exactly 92.5000% of par, for proceeds of 938.872 bn. One auction supplied 23.5% of everything Minfin placed at nominal this year.
Around it the record is thinner than the total suggests. From 17 June until the 2 September auction Minfin placed 10.4 bn roubles, all of it on 1 July. On 15 July it published this:
“The Ministry of Finance of the Russian Federation reports that the auction to place issue No. 29028RMFS of federal loan bonds with a floating coupon, held on 15 July 2026, was declared void owing to the absence of bids at acceptable price levels.”
The table records 144.3 bn of demand at that auction and nothing placed. On 4 February the same thing happened to issue 26251RMFS with 294.0 bn of demand. In both cases buyers were present and Minfin refused their price — which is the same refusal that produced the freeze we described in August, and consistent with what we found when banks cut their OFZ holdings in July. What an OFZ is, and why a failed auction cannot be hidden, we explain separately.
What this changes in our model
Nothing, and by design. OFZ placement was retired as an indicator on 22 August because it measures how the deficit is financed rather than whether Russia can pay for the war, and because it duplicated the liquid NWF we already measure directly. This is a note about a mechanism, not an indicator update.
The budget deficit scores 65 out of 100, with the rolling twelve-month deficit at 3.3% of GDP. The 450.0 bn gap is not itself a financing shortfall: the cash side is what the budget always received, and the programme was always written in nominal. It is the size of the wedge between the two ways of reporting the same year — and the amount by which a reader who follows the headline overstates what borrowing has delivered.
The strength index stands at 47 out of 100, in the stress zone. No indicator moved in today's data update.
What we do not know
The workbook is current to 3 September, so the 9 September results are not in it. Our year figures are that table plus the three notices published on 9 September, and they will have to be rechecked when the file is next updated.
The auction notices give price and yield but not the coupon of each issue, so we do not say how much of any discount is a low coupon and how much is anything else. Minfin has not published a borrowing schedule for the fourth quarter. And beyond “the absence of bids at acceptable price levels”, the 15 July notice gives no reason why 144.3 bn of demand produced no placement.