Is this structural or cyclical
Crude capacity churns: it is hit, repaired in weeks, and comes back. The high-octane chain does not, because reforming has no proven domestic substitute at the modern end and parts arrive at a production rate rather than a lead time. When these two series separate and stay separated, the shortage is structural.
Who is fed, and at what level
Tiers 1 and 2 are absolute claims: the state serves them first and in full while anything is left. Tiers 3 and 4 share the remainder with ordinary retail, pro rata. Ladder, widget 1.3.
Regions
Provisional demand shares. Product moves only between adjacent regions, at most 70% of a region's supply leaves it, and only Moscow may draw from anywhere. Volume is conserved: what the shield keeps, other regions lose.
Repair chain, spares and maintenance
A hit is not repaired from a warehouse. Each unit class has a national pool of spare sets; once it is empty, the next repair waits for manufacture and grey import. Deferred turnarounds raise the chance of unplanned stops until the winter wave runs them.
What moves the number
Ledger for the selected horizon, kt per week
Beyond repair at this horizon
How much rests on the numbers with no source
Five constants in this model are judgments rather than measurements. This shows how much of the twelve-month answer each one owns, by moving it up and down by half. Anything with a wide bar is a number a reviewer should be asked about before the result is quoted.
Who pays
Starting state
Refinery registry
Anchored constants
Movable only inside a published range. Constants marked unsourced are judgments, not measurements. Override them through the scenario JSON with a source; the override is listed in the passport.
The Direct Receipt
Computed from the ledger this scenario produces, cumulative to each date, in billions of roubles. Every line is reproducible from the ledger CSV. This is only the part that is easy to calculate.
What a driver actually pays on the street
The official price is administered, so the shortage shows up as a street premium rather than as an official price rise. That premium is not in the consumer price index but it is in household experience, which is what drives expectations.
The parallel-market premium is assumed to rise convexly as the physically available share falls, because discretionary demand exits first and the residual market increasingly consists of high-value trips. The upper range is capped at four times the administered price, deliberately below the five to seven times that affluent households could sustain after sharply reducing consumption. Even at the top of the band, some poorer households would still buy for necessary journeys, to a hospital or once a week for food, while cutting everything else. This parameter is a modelling assumption, not an empirical elasticity estimate.
Second-order consequences
Order-of-magnitude bands intended to show relative size, not estimates. Where a number could be computed it is in the receipt above; everything here is deliberately left in bands.
Scenario passport
Everything a reader needs to reproduce or dispute this scenario. It travels with every export.