On 28 October the Bank of Russia published the final text of its monetary policy guidelines for 2027–2029, at the same address where the September draft had stood for four weeks. The bank’s own list of changes describes what happened to the box on budget policy in one line: the text of box 9 was “adjusted to take account of current budget-policy parameters”.

We hold both files. The draft is 227 pages, its cover reads “Draft of 30 September 2026”, and it was “approved by the Board of Directors of the Bank of Russia on 29.09.2026”. The final is 251 pages and its cover no longer says “draft”: “approved by the Board of Directors of the Bank of Russia on 28.10.2026”, prepared on statistics to 20.10.2026, with a forecast data cut-off of 22.10.2026. Read side by side, they show changes the change list does not mention.

The premise changed sides

The largest of them is not in box 9 at all. It is in the description of the baseline scenario, and it is the sentence that says what the bank’s central forecast rests on.

In the draft:

“The assumptions of the baseline scenario regarding budget policy rest on public comments by representatives of Russia’s Minfin about a gradual reduction of the federal budget’s structural primary deficit with its exit to a zero level in 2029. The forecast calculations incorporate a trajectory assuming a structural primary deficit of 2% of GDP in 2026, 1% of GDP in 2027, 0.5% of GDP in 2028. At the same time, in the baseline scenario the Bank of Russia proceeds from the base price currently in force in the budget rule, established by the Budget Code of the Russian Federation.”

In the final text, the same passage reads:

“The assumptions of the baseline scenario regarding budget policy rest on the draft federal law on the federal budget submitted to the State Duma and the Guidelines for budget, tax and customs-tariff policy of the Russian Federation for 2027 and the planning period 2028 and 2029. They assume a reduction of the base oil price in the budget rule to 50 US dollars per barrel in 2027–2029, as well as a gradual reduction of the federal budget’s structural primary deficit to a zero level in 2029.”

Two things went out of that paragraph. The bank’s own numbers — 2%, 1%, 0.5% — and the clause saying it was working from the base price currently in force. The phrase “the base price currently in force” appears nowhere in the 251 pages of the final document.

The introduction moved the same way. The draft told the reader the bank “proceeds from a gradual reduction of the structural primary deficit from 2% of GDP in 2026 to 1% of GDP in 2027 and 0.5% of GDP in 2028, reaching a zero level in 2029” and that “preservation of the base oil price currently in force in the budget rule is also assumed”. The final says the bank proceeds “from the new budget projections, assuming a gradual reduction of the structural primary deficit from 1.8% of GDP in 2026 to 0.6% of GDP in 2027 and 0.2% of GDP in 2028, reaching a zero level in 2029”, and that “a reduction of the base oil price in the budget rule to 50 US dollars per barrel from 2027 is also planned”.

A month ago the central bank forecast on the law. It now forecasts on the bill — and says so in its own words, because the sentence names what it rests on: a draft federal law submitted to the State Duma. The base price it assumes for 2027 is $50. The schedule actually written in point 4 of article 96⁶ of the Budget Code, as amended by law No. 432-FZ of 28 November 2025, which we read in full on 1 October, is $58 for 2027. Changing it needs the Budget Code amendment tabled as order 2707-r on 30 September, whose text has never been published and whose passage we cannot confirm: sozd.duma.gov.ru and duma.gov.ru have been unreachable for a fourth day.

This is not a quibble about drafting. The guidelines are the document in which Minfin’s projections are restated in another institution’s hand and become the premise of a macroeconomic forecast and of the key rate — held at 14.00% from 26 October by the decision of 23 October that this edition incorporates. A premise taken from a bill is a premise that fails if the bill changes.

The paragraph that is no longer there

Box 9 ended, in the draft, with three sentences measuring the new plans against two benchmarks — the bank’s own July forecast, and the plans as they stood before this year’s revision:

“According to the new budget projections the trajectory of the structural primary deficit over the coming two years is somewhat below the assumptions of the Bank of Russia’s July baseline forecast. All else equal this means a somewhat smaller contribution of budget policy to aggregate demand than was assumed in July. At the same time this contribution in 2026–2028 remains more significant than it was in the budget plans before their revision in 2026, which provided for a return to a zero structural primary deficit already this year.”

That paragraph is gone. In its place the final text has:

“Under their implementation budget policy will exert a disinflationary influence over the forecast horizon. At the same time, in 2026–2028 the structural primary deficit will persist, although it will gradually contract. This means that budget policy will continue to stimulate aggregate demand. Given limited possibilities for expanding supply, maintaining balance in the economy will require more restrained dynamics of other components of domestic demand.”

The replacement is not a softer version of the same claim. It is a different claim. The deleted sentences compared this plan with an earlier plan and found this one looser; the new ones describe the plan against no benchmark at all and open with the word disinflationary. The one place in the document where the reader was told that the zero date had moved — that budget plans had once provided for a zero structural primary deficit “already this year” — is the place that was removed.

We quoted that sentence in our piece of 1 October, from the draft, and the quotation stands: it was there, and we can show the file it was in.

What got milder, and what did not move

The “Risk” scenario is the bank’s worst case — a collapse in commodity prices and a flight from technology assets. Its language about the fund is unchanged between the two editions, word for word:

“against the background of a significant fall in commodity prices for oil, under the current parameters of the budget rule this will lead to intensive use of the liquid part of the NWF, which creates risks of a rapid exhaustion of the fund’s resources”

What changed is the price at which the bank says the rule would then have to be rebuilt. The draft required “a gradual transition to new levels of the base oil price — 40 US dollars per barrel in 2027 and 35 US dollars per barrel from 2028”. The final says 45 dollars in 2027, with 35 from 2028 unchanged. The worst case got five dollars cheaper for one year.

Two other changes in box 9 are small and both point the same way. The structural-deficit sentence gained its 2026 figure, “from 1.8% of GDP in 2026” — a number Minfin had already published itself three weeks earlier, in the January–September execution report we wrote about on 10 October, where the deviation from the structural target was admitted in the ministry’s own words. And the list of sources for higher non-oil-and-gas revenue gained the profit tax: the draft credited “a widening of the tax base under the progressive personal income tax scale”, the final “a widening of the tax base under the profit tax and the progressive personal income tax scale”.

One sentence that did not change is worth naming, because it is easy to read as new and it is not:

“According to the estimate of Russia’s Minfin, the liquid part of the NWF will be replenished across the whole medium-term horizon.”

That stands in both editions, in the same place, in the same words. It was in the draft of 30 September and we quoted it on 1 October. The tension it creates with the risk scenario on the same document’s other pages is real, and it is a month old, not a day old.

Why this is about the ability to pay for the war

The structural primary deficit is the measure that strips out oil and gas price swings and interest, and it is therefore the closest thing to a published statement of how much the budget is spending beyond its means for reasons of its own. Its path is the spine of both versions of this document, and between 30 September and 28 October it was rewritten twice over: the numbers changed from the bank’s to the ministry’s, and the comparison that showed the plan had loosened was taken out.

What remains is a central bank whose baseline now assumes a budget that has not been passed, a base oil price that is not in law, and a fund its own worst case says could empty quickly while the ministry’s estimate in the same box says it will be topped up throughout. None of those are our judgements about the document; each is a sentence in it.

The practical effect is on what can be checked. When the forecast rested on the Budget Code, a reader could open the code and see the number. Now the premise is a bill nobody outside has read, in a legislature whose own systems have been dark for four days — and the one paragraph that measured the new plan against the old one is not in the file.

What this changes in our model

Nothing today. Not one of the eight readings moved: the parser’s run this morning recorded zero indicator changes, and neither edition of the guidelines contains an observation we take.

The series this story runs through are the budget deficit, at 66 of 100, a rolling twelve-month deficit of 3.2% of GDP against a pre-war anchor that was a surplus of 0.34%; and the liquid part of the NWF, at 27 of 100, about $45.0 billion at the end of September against $113.5 billion before the invasion. Oil and gas revenues stand at 10 of 100, and neither edition moves them: the indicator measures the level of revenue actually collected, not the base price that decides how much of it may be spent. The strength index is 46 of 100, in the stress zone; all eight readings are on the monitor.

A forecast premise is not a reading and never becomes one. What it changes is the frame in which the readings are published, which is why we record it and do not score it.

What we do not know

Whether the Budget Code amendment has passed. We have never seen its text. It was tabled on 30 September as order 2707-r; the Duma’s bill system and main site have been unreachable for four days, and the first reading of the budget package was due on 29 October. The bank’s own sentence calls it a draft law submitted to the State Duma, which is the strongest statement available to us and is as of 28 October.

Why the paragraph went. The change list gives a reason for the box — current budget-policy parameters — and no reason for any individual sentence. We do not assert an intention and did not find one stated.

What else moved in 251 pages. We read the introduction, box 9, the baseline-scenario assumptions, the “Risk” scenario and the change list against their counterparts in the draft. The final edition also adds box 14 on companies’ interest costs and a whole appendix 8 on inflation targeting across countries, including box 16 on unconventional measures. We have not read those against anything, because there is nothing in the draft to read them against.

What Minfin’s estimate rests on. The replenishment of the liquid part of the fund “across the whole medium-term horizon” is attributed to Minfin in both editions and sourced to no document in either. minfin.gov.ru has returned 503 for a fourth day.