Economics
Energy
Finance

Seven mistakes in analyzing data from Russia — and how to avoid them

Dmytro Voronenko on how official Russian numbers are “rebuilt”

Читать на русском
Photo: Scanpix

The Kremlin behaves as if it knows which Russian figures Western think tanks and governments actually use, and how they use them. Those figures then disappear, are rebuilt on new rules, or are “clarified” until they no longer mean what outsiders think they mean. Hiding the effect of sanctions and the real state of the economy is part of a wider information war: abroad, to steer decisions Moscow wants; at home, to hide how the cost of the war is being pushed onto firms and households. In a long war, this is one of the main places the Kremlin spends its effort.

The same patterns appear in work on indicators of Russia’s ability to last in a war of attrition, and just as often in everyday commentary on the economy. They come more easily to analysts from democracies, who struggle to picture how an authoritarian government can massage official numbers, lean on administrative — and often unofficial — measures, or quietly tap the non-state part of the economy.

Most examples below come from oil and fuel. Ukrainian strikes have made that the live subject, and enough of the relevant decisions are public to show how the trick works from start to finish. Some cases are detailed on purpose: manipulation is rarely crude; it sits inside procedure. The list is not ranked and not complete.

1. Treating official figures as market prices when the state can set them

Textbooks favor headline figures — inflation, GDP, output — because they are widely cited and appear hard to dispute. That makes them a primary target for manipulation. Annual revision of the inflation basket is normal; the direction of Rosstat’s 2026 revision is not. Broader item specifications introduced in 2026 — such as omitting fabric types for clothing or tracking medicines by generic names — create room for substitution bias by allowing cheaper alternatives to be recorded without quality adjustment. Since private consumption accounts for a major share of GDP, systematic distortions in consumer price measurement inevitably leak into the GDP deflator, making the official “real” GDP growth rate equally questionable.

Oil output is another case. Crude, gasoline and diesel production figures have been withdrawn from public release. Prices still appear every week and by grade, which makes the statistical record look intact. Production numbers then leak through “sources” citing closed CDU TEK bulletins, or are reverse-engineered from the mineral extraction tax. In both cases the number is reconstructed from a source inside the industry, using parameters the state itself sets: tax rates, coefficients, deductions. Tax receipts are reliable; the physical volume inferred from them is only as reliable as those coefficients.

Say which part of the figure is set by the state — which decision, from which date, and how it moves the number.

2. Repeating another shop’s numbers without their warnings — or their mistakes.

Tell readers where the number came from. The washing often works like this. The IMF takes a Rosstat figure, applies a light tweak, and publishes a derived indicator. A research center treats the IMF number as raw material, runs a standard method, and produces a “neutral” estimate that everyone else cites. Whether it means to or not, it has cleaned data shaped for the Kremlin.

Real GDP is the cleanest example. It is calculated by deflating individual components of national accounts using specific price indices. These indices come from the same price surveys, by the same agency (Rosstat), whose inflation figure is already in dispute — and which also publishes the nominal total. That is not an independent check. It is circular. But the figure then travels through international organizations and comes back looking clean. Nobody in the chain is doing anything improper. The IMF publishes what member states report; that is its mandate. The research center applies a standard method to a published international figure. The laundering is a property of the chain — which is why nobody stops it.

Kpler is the other common case. The firm publishes what tanker- and satellite-tracking can do: the best picture available, not a full census. That warning then drops out. Analysts quote the numbers as settled fact, and Kpler’s reputation does the work the warning was meant to do. Passing the caveat on is the analyst’s job, not Kpler’s. For Kpler the series is one product among thousands; for the Russia specialist it is the whole subject.

Trace every number to its first source and put that source’s limits beside the figure, not in a footnote.

A close cousin is the even-handed haircut. If a source is discounted for bias, the discount should have a direction. A symmetrical range around a one-way distortion leaves that center in place and merely makes it look processed.

3. Ignoring the decrees that changed what the number means

There is no excuse for not knowing which official decisions have altered the indicator you are using.

The central bank’s key rate is no longer mainly a tool for pricing credit or holding down inflation. On 24 July 2026 the Bank of Russia cut the rate to 14.00 per cent — the fifth cut of the year — on the same day it raised its inflation forecast from 4.5−5.5 per cent to 6−7 per cent. Whatever the rate is being set for, it is no longer only the forecast.

The share of gasoline that must be sold on the exchange is another case. On paper the rule is still 10 per cent, not 2 per cent. The mandated share was set at 10 per cent in 2013, raised to 13 and then 15 per cent in 2023, then cut back to 10 per cent in July 2026 — which can look like a modest step down from “the usual 15 per cent.” Resolution No. 1003 of 10 August 2026 then “clarified” compliance: up to 8 per cent can be counted through confirmed off-exchange deliveries to farms, Russian Railways, builders, or anyone collecting the fuel themselves. Only 2 per cent must still be sold freely on the exchange. Leave the headline in place, change the small print, and mislead anyone who reads only the headline.

Exchange data still work as a pair: listed volume against the street price. Real scarcity moves three parameters together: volume down, exchange price up, informal premium up. Manufactured supply looks different: volume steady, exchange price soft, informal premium rising, availability at the pump falling. The diagnostic is the alignment of the signs.

Pump prices are managed twice over. Big oil companies get direct orders and a damper subsidy — extra money paid only while the exchange price stays close to a state-set reference. Independent chains, with no refinery and no damper, are kept in line by the antimonopoly service: warnings, dominance cases, fines on turnover, and an unwritten benchmark of “no faster than inflation.” The benchmark does not bind the outcome — pump prices have outpaced inflation, especially at independents — but it binds behavior. Sellers who cannot price freely compensate on availability and grade, which is why the shortage shows up as missing 95-octane rather than a bigger price spike. The only price that still tells you something is the black-market price, and there is no respectable public series for it. Exchange prices and pump prices at the majors are policy tools; pricing at independents measures survival, not demand.

A published figure is not ready to use until the last year of official decisions has been checked.

4. Using a broad total when the part that matters is missing

This also covers failing to notice when the authorities add a new ingredient that keeps the total looking healthy.

Guessing a missing piece from its usual share of the total is reasonable when no separate figure exists — especially in a democracy. It is a mistake when that piece matters to an authoritarian government at war.

Refinery throughput is not high-octane gasoline. Three facts get lost. First, basic crude distillation is simpler, mostly made at home, and easier to repair; the secondary units that upgrade fuel stay down longest. The plant cannot just stop, or wells would have to be shut in, so crude keeps moving through whatever still works and comes out as straight-run gasoline, naphtha, diesel or fuel oil. The mix shifts; the share of sellable gasoline falls.

Second, fuel standards have been loosened — Euro 3 in July, Euro 2 at the pump from 1 September 2026 — so the old split inside “gasoline” no longer holds. Drivers will pay later in engine repairs: a quiet tax on keeping the war going. Third, many analysts never separate 95, 98 and 100 octane, where the shortage is sharpest. They report “refining” instead. The grade people need is often not for sale; on the spreadsheet there is only a modest shortfall.

Track the component itself. If no such figure exists, say so and publish an estimate with the assumptions in view. Do not infer it from an average share that no longer fits.

5. Writing as if the last four years had not happened

Any reading of the numbers now sits in a different setting: more than four and a half years of war, which have used up easy ways of adapting and hiding damage and have run down many reserves; distortions already baked into the statistics; no access to foreign borrowing; perceived inflation of 15.1 per cent as measured by inFOM; and self-censorship among firms and ordinary people.

The refinery strikes of August 2026 are not those of 2024, or even of autumn 2025. This is a third wave of concentrated attacks. Spare-part kits are gone, especially for units that make high-octane gasoline. Plants take longer to come back. In 2025 a fifth of refining capacity was offline at the peak, yet total output fell only 6 per cent — spare units covered it. Now refining is at a 21-year low, and even the Moscow refinery, hit twice in three days in June, is not expected fully back before 2027. Governors now talk about long repairs. One fact cuts the other way: basic distillation is made at home and does get repaired, so the headline “refining” number will look better than high-octane output. That gap is what gives an analyst who reads only the total the impression of recovery.

The August strikes only make sense against that backdrop, and the backdrop has to be dated as carefully as the production figures themselves.

6. Forgetting the stocks and delays Moscow uses to buy time

A buffer is a stockpile, a reshuffle inside the system, or an official delay that keeps the published number from showing the real damage for a while. It does not undo the damage. It postpones the moment it becomes visible. A conclusion that “so far everything is fine,” read off a buffered figure, is a statement about how much stock is left, not about the health of the system.

In earlier gasoline crunches the Kremlin drew down stocks at refineries, depots and filling stations to hide how much capacity had been hit. In this third wave there is much less slack. A calm picture may be the buffer running down. It is still too soon to judge the damage by what is on sale at the pump. A buffer is still there. It is thinner than it was.

The published number is useful only with an estimate of what buffer remains — not as a substitute for one.

7. Treating a rescue measure as proof the state is in control

The analyst sees a support measure and reads it as strength: the state has stepped in, so the line is holding. Each such measure is a dated admission that the line has already moved. Stopping publication is the same thing. When a figure Moscow used to release turns visibly bad — cash on the treasury’s accounts is the obvious candidate — the series goes dark. Analysts then drop it, instead of treating the blackout as the message. Hiding it now costs less than showing it, which puts the true value below anything the last printed point would support. This is not the most common mistake. It is the most damaging, because it turns evidence into proof of the opposite.

The fuel market is a sequence of those admissions: the first-ever cut in the mandated exchange share of gasoline; the decree of 12 October 2025, which removed the wholesale-price test from the damper subsidy until 1 May 2026; the later widening of the allowed gap for gasoline from 10 to 20 per cent; and, on 31 August 2026, permission to sell Euro-2 fuel at the pump, billed as a way to “stabilize supply.” That is not stabilization. It is the official formula giving way, one rewrite at a time.

Top reads
  • Why Russia Will Not Attack NATO
  • The Kremlin’s veteran hero cult stops at the workplace door
  • Germany and Russia: From “Change Through Trade” to Sharp Power
  • Russia and Belarus: A Satellite with Room to Maneuver
  • Managed Elections and Unmanageable Interest
  • Turkey and Russia: Economic Rapprochement amid Political Rivalry

It is getting more and more difficult for independent analysis to survive in today’s conditions. We at Riddle remain committed to keeping all our texts freely available. So paywall subscriptions are not an option. Nor do we take money that may compromise the independence of our editorial policy. So we feel forced to ask our readers for help. Your support will enable us to keep on doing what we believe in, without fear or favour.

Read also
Russian Economy After the Elections

Vladislav Inozemtsev on why the Kremlin is likely to favor control and expropriation over putting the economy on a war footing after the State Duma elections

Burn Warehouses, Crash Banks?

Oleg Loginov on the fallout from attacks on Wildberries, "Russia's Amazon"

Prodding Open Wounds: Ukraine’s War on the Russian Oil & Gas Sector

Nicholas Trickett on how Ukraine’s strikes on Russian refineries are deepening the contradictions of the Kremlin’s wartime economy

Search