How long can Russia pay for its war?
Who pays now, and who pays for the next move.

Standard measures make Russian public finances look comfortable. Gross financing need for the federal budget in 2026 comes to 20–25% of revenue, which is nothing alarming by international standards. But the federal budget is only one accounting perimeter, while the obligation sits on a system. A region that cannot pay its contractor, a defence plant whose bank declines to roll over working capital, an unpaid salary: three different routes to the same failure, all funded out of the same rubles.

Count it that way and the 2026 requirement is RUB 16.1 trn (USD 196 bn). That is 40% of federal revenue, and 124% of the National Defence line: more than the whole budget item. Working range RUB 13.1–18.9 trn (USD 160–230 bn, converted at 82 RUB/USD). Composition and double-count checks are published.

The requirement is still being met. What this panel tracks is how, and at whose expense. Accumulated liquidity first, then voluntary purchase, now compulsory absorption and central bank refinancing. Wages indexed below the inflation people actually see. Anti-drone defence pushed onto private balance sheets. Regional budgets. Bank capital. Depositors.

The state decides where the cost is booked. Practice reveals who ultimately pays. This panel tracks the second, because that is the capacity that runs out.

It will not tell you when the war stops, since the stopping point is a decision, not an observable variable. It measures the constraints that shape that decision instead: how much room remains on each ladder, who pays for the next move, and what the last ones cost.

Every rung marked as passed is pinned to an observable event: a decision, a payment, an administrative measure, a physical failure. Thresholds are chosen by rule, not adjusted to make the current reading work. The order of the rungs is a judgment and is labelled as one. Where the panel makes a forward-looking assessment, the falsification condition and the check date are published in advance. Figures from interested parties are used; their adjectives are not. Anything that cuts against the argument goes first.

Updates

Panel revisions and reviews of expired check dates are published at Incidence Notes — or via RSS, if you would rather not leave an address.

Acknowledgment

My thanks and my respect go to the people who did the intellectually challenging work this panel rests on, under conditions that often make an assessment, modelling, discovery of hidden aspects or confirmation of facts expensive: analysts at think tanks and specialist research firms, OSINT researchers and the verification communities around them, journalists, academic institutions, and independent researchers.

Although they are not one community, and all of them work separately, to different standards, and frequently disagree with one another and with this panel, the acknowledgment goes wholeheartedly to their work.