Calibrating a scale on administrative action

The natural indicator is the shortfall: −35%, against a threshold of −20%, or −25%, or whatever the analyst picks. Three things are wrong with it, whether the analyst notices or not: 1) the threshold has no defensible origin, since any number invites the question of why it is not five points lower and the only answer is the analyst’s judgement, which is a weak thing to put underneath the reading; 2) the scale saturates, since once the shortfall is past the threshold the indicator has said everything it can say, while the interesting movement happens afterwards — a state facing a 35% gap does things it was not doing at 20%, and those things are the reading; 3) the series is soft exactly where it matters, since output, demand and coverage are estimates with wide intervals, published by parties with an interest in them, and revisable.

A regional decree either exists or it does not, and it carries a date. The administrative record is harder than the quantitative one.

Seven rungs, each defined by an action taken by an authority or a seller, each observable from a decree or from a notice taped to a pump. Zones attach to rungs, not to percentages.

The rungs are not four variants of one thing called “rationing”. Rungs 2 and 4 cap the volume of a single purchase and the buyer comes back in the evening; rung 5 caps the frequency of access, which is a refusal to sell to a particular person on a particular day; rung 6 introduces personalised accounting and converts petrol from a good into an allocated resource = the buyer stops being a customer and becomes a recipient. Collapse the four into one category and the only movement worth watching disappears.

Rung 4 was reached in June 2026, on the first wave of strikes, at a fall in domestic supply of about 20%. That figure is recorded as the point at which the scale was calibrated: a date and an observed correspondence, not a judgement. Thus “why 20 and not 15” has an answer that does not belong to the analyst — 20 is where the state acted.

Two consequences. The current state is worse than the calibration point — coverage of about 65% of seasonal demand, a gap near 35% — and the scale keeps working through that deterioration, since rungs 5, 6 and 7 lie above it. And the calibration is checkable by anyone: it fails if the correspondence between shortfall and response does not hold on the next episode.

A percentage indicator cannot record this at all: the shortfall series shows a small wobble, or nothing. The ladder records a failed relief — precisely the observation that would have refuted the panel’s standing claim had it held — and logs it as a candidate that did not materialise.

The scale also names the payer at each rung, which a shortfall figure never does. Rung 3 is the clearest case: permitting fuel of a lower emissions class closes the shortage by quality instead of by volume, and the bill goes to the car owner in engine wear = a tacit tax on continuing the war, collected in repairs rather than in roubles. Introduced by resolution No. 819 of 2 July 2026 and widened on 5 August; widening it five weeks later means the first wave solved nothing.

Diesel demand is compressed by decision, through administered consumers: the shortage is absorbed by prioritising deliveries by category and stays invisible. Petrol demand is atomised across millions of buyers who cannot be prioritised quietly, thus rationing petrol is a public demonstration of incapacity rather than a regulatory measure. The export cushion runs on the same axis — 10–15% of output for petrol against about 50% for diesel — so an export ban buys weeks on one and quarters on the other.

Rule: pick the good whose shortage the state cannot allocate out of sight. An indicator built on the administered good measures the administration, not the shortage.

• Seasonality is not eliminated, it is dated. The amplitude between summer peak and winter trough runs 10–15%, and that is all the series will support. Rung 4 or higher persisting through December–February removes the explanation entirely; until then it stays available to a critic, and the widget says so.
• The ordering of rungs is a judgement. Observability applies to each rung; the sequence between them rests on which restriction is the heavier imposition on the buyer. Argued, not measured, and open to reordering on a reasoned proposal.
• Partial passage does not move the zone. Rung 5 operates in seven regions and rung 7 appears in the justification of one operational headquarters, and neither is credited until it holds in full, thus the ladder reads late by design.
• Refutation: volume caps lifted in most regions, domestic supply back at last year’s level, free export resumed. One observation, three components, all public.