On 2 September the CBR published the September round of its macroeconomic survey. Thirty-three economists take part; thirty answered the budget question. Their median puts the 2026 consolidated budget deficit at 3.5% of GDP, against 3.2% in the July round. The forecasts for 2027 and 2028 also worsened. In the same round, the panel forecast a weaker rouble across the whole horizon — a move usually described as helping Russia's budget.
What the survey says
The document is “Macroeconomic Survey of the Bank of Russia. Survey results: September 2026”. It was published on the bank's statistics pages on 2 September as a PDF. Fieldwork ran from 28 August to 1 September. Figures in brackets throughout the tables are from the July round, so both rounds can be read from one file.
“Consolidated budget balance: Analysts expect a larger consolidated budget deficit in 2026–2028 — 3.5% of GDP (larger by 0.3 pp) in 2026, 2.6% of GDP (larger by 0.1 pp) in 2027, 2.1% of GDP (larger by 0.2 pp) in 2028. The 2029 deficit forecast is lowered by 0.1 pp to 1.4% of GDP.”
Three of the four forecast years got worse. The fourth, the most distant, improved by the smallest increment the survey records.
Thirty-three economists, thirty answers
The methodology note is explicit about what the median is:
“Method of calculation: the survey results are the median of the forecasts of 33 economists from various organisations taking part in the survey. If a respondent gave their expectations as a range, its midpoint was used in the calculation.”
Thirty-three is the panel size. The bank also publishes the full set of answers as a workbook alongside the PDF. It gives the count for each question: the 2026 consolidated-budget line has 30 answers this round, against 28 in July. The headline figure is the median of thirty forecasts, not thirty-three.
The rouble moved in the budget's favour
In the same round the panel forecast a weaker currency:
“USD/RUB rate: Analysts expect a weaker rouble (by 1.4–2.5%) across the whole horizon compared with the July survey. The forecast for 2026 is 79.9 roubles per dollar (implying an average rate of 84.9 roubles per dollar in September–December 2026), for 2027 88.8 roubles per dollar, for 2028 94.0 roubles per dollar, for 2029 97.2 roubles per dollar.”
The panel also raised its 2026 forecast for the oil price in dollars. The survey does not ask about Brent; it asks about the price to which budget revenue is actually tied. Footnote 2 defines it as “the price of Russian oil determined for tax purposes and published monthly on the official site of Russia's Ministry of Economic Development”. For 2026, that price rose from 63 dollars a barrel in the July round to 64 dollars now.
A weaker currency and a higher dollar oil price are the two halves of the standard explanation for why Russia's budget keeps balancing: oil is sold in dollars, while wages and procurement are paid in roubles. Both moved in the budget's favour in this round. Multiply the panel's two medians for 2026 and a barrel is worth 5,113.6 roubles, against 4,939.2 in July — about 3.5% more roubles per barrel.
That multiplication is ours, and it is an illustration rather than a forecast. Each figure is the median of a separate distribution, and the median of a product is not the product of the medians; the survey never asks anyone to forecast a rouble price per barrel. It is enough to show the direction: in the same document, the panel raised the rouble value of a barrel of oil and worsened the deficit forecast.
The denominator did not do it either
A deficit expressed as a share of GDP can worsen because the numerator grew or because the denominator shrank. The survey publishes no nominal GDP and never asks for a deflator, so the denominator cannot be read off this file. It can be bounded. For the denominator alone to carry the whole revision, nominal GDP for 2026 would have to come out about 8.6% below what the July round implied — that is what it takes to move an unchanged rouble deficit from 3.2% of GDP to 3.5%.
Nothing the survey publishes is near that size. Real growth for 2026 was cut by a tenth of a point, from 0.6% to 0.5%. Average-year inflation was raised by a tenth of a point, from 6.0% to 6.1% — and faster prices make nominal GDP larger, which pushes the ratio the other way. The CPI is not the GDP deflator and we do not treat it as one. The claim here is only about magnitude: the revision is 0.3 pp, and the two published moves that bear on the denominator are 0.1 pp each, one of them with the wrong sign.
Better than last year, worse than it was going to be
The direction matters more than the level here, which is exactly why the level should be stated plainly. The table's fact row puts the 2025 consolidated balance at −3.9% of GDP. A 2026 deficit of 3.5% is an improvement on last year, and the panel is not forecasting a collapse. Between July and September, these economists did not start expecting a bad year. They began expecting a smaller improvement than they had two months earlier.
The longer view sits in the same row. In 2021 the consolidated budget ran a surplus of 0.8% of GDP. The panel's horizon ends in 2029 at a deficit of 1.4%. On these forecasts the budget does not return to where it was before the war at any point the survey covers.
The export line tells the same story with a different measure:
“Exports of goods and services: Analysts lowered the 2026 forecast slightly and raised the later years — 508 bn dollars (−2 bn dollars) in 2026, 486 bn dollars (+3 bn dollars) in 2027 and 495 bn dollars (+6 bn dollars) in 2028. The 2029 forecast is 501 bn dollars (+4 bn dollars). That is 9% (49 bn dollars) below exports in 2021.”
The comparison with 2021 is the bank's own, made in its own document. It is not our oil-and-gas indicator, which measures budget revenue in roubles rather than exports in dollars. But it applies the same test: measure against the last pre-war year, not last year.
What the spread does and does not show
The workbook also shows the distribution of answers. For 2026, the panel's most optimistic forecast is a deficit of 1.7% of GDP — exactly the same as in July. The most pessimistic is 6.1% of GDP, against 4.2% in July. The mean moved further than the median: from 3.1% to 3.5%, while the median went from 3.2% to 3.5%. That is what a lengthening tail on one side does to an average.
Two cautions belong beside that. The band from the 10th to the 90th percentile did not widen: it ran from 2.31% to 3.79% in July and from 2.71% to 4.00% now. The middle of the distribution therefore shifted towards larger deficits while narrowing slightly. The bank says as much in its summary: the ranges narrowed for most indicators. The number of answers to this question also rose from 28 to 30, so the new extreme may come from a new respondent rather than a changed view. One anonymous forecast of 6.1% is not a finding. It is the edge of a distribution, and we cannot attribute it to anyone.
What this changes in our model
Nothing, and the reason is worth stating precisely. Our federal budget deficit indicator calculates the federal budget's rolling twelve-month deficit as a share of GDP from Minfin's monthly reporting: 7,460 bn roubles through July, or 3.3% of GDP. The survey covers the broader consolidated budget, including regional budgets, and gives an annual forecast rather than an observed rolling window. The document names the series but nowhere defines its perimeter.
So we do not put 3.5% beside 3.3% and treat one as a check on the other. Nor do we add our regional balance — a deficit of 1.616 trn roubles over twelve months — to the federal figure to manufacture a comparison: transfers between the two budgets are netted out in any consolidation, so the sum of our two numbers is not a consolidated anything. The same rule applied on 1 September to the CBR's own Monetary Policy Guidelines, where a structural primary deficit of 2% of GDP could not be set beside the parser's 3.3% either.
The connection to that piece is worth drawing anyway. The Guidelines were published on 31 August with a fiscal path the bank had to assume because the government has not set one. The survey followed on 2 September, with the bank's panel of outside analysts downgrading the budget outlook. Four days, two documents, one central bank — and neither contains a government statement of what the path will be.
One more line points in the same direction as the Guidelines. The panel left its 2026 key rate forecast at 14.5% and raised the 2027 forecast from 12.2% to 12.4%, in the same round in which it worsened the 2027 deficit forecast. That is the direction the bank's own document described: delayed consolidation means less room to ease. A single round in which two numbers move together is not evidence of the mechanism. The panel's numbers merely do not contradict it.
What we do not know
The survey publishes a median, a mean, percentiles and a count, and nothing that attaches a forecast to a name — the full array of answers carries no organisations. There is no split of the consolidated balance into its federal and regional parts, which is the gap our regional indicator exists to fill from other data. There is no nominal GDP, so the rouble deficit these percentages imply cannot be recovered from the file. And the document never defines what its consolidated budget contains, so whether the state extrabudgetary funds sit inside that perimeter is not answerable from the document. The next round of the survey runs from 9 to 13 October.