# Goods trade with Russia and Ukraine

Goods trade is the largest part of the map for most countries: what each country bought from and sold to Russia and Ukraine, 2019 to August 2026. This page explains where the numbers come from, how value is moved to the country where it really belongs (re-export hubs, transit, gold), and why sales count as profit rather than revenue. Fuel-specific steps are on the [energy page](https://warbalance.org/method/energy-and-laundered-oil/).

Details for developers: [filling order, shares, margins and research files](https://github.com/intelligent-cream/war-balance/blob/main/docs/reference/method/trade.md).

## Sources and filling order

233 economies, four flows (bought from and sold to Russia and Ukraine). Values are taken in this order:

1. the country's own customs report (UN Comtrade; Eurostat for EU members in 2026), certainty A;
2. the partner's report of the same trade (Russia's or Ukraine's), adjusted for the usual gap between the two, certainty B;
3. IMF trade statistics, certainty B;
4. an estimate from the growth of reporting countries, certainty C.

Russia publishes no partner detail after 2021, so from 2022 trade with Russia is always what the other country reports. Official national statistics replace these where they are better (Belarus, Kazakhstan, Kyrgyzstan, Armenia, Uzbekistan, Türkiye, the UAE and others). Kyrgyzstan's 2020 and 2024 to August 2026 are its statistics office's revised totals (Comtrade holds an earlier, lower version of the same figures), and the estimate of its unrecorded sales to Russia is cut by the extra sales now reported, so they are not counted twice; Armenia's January–August 2026 is Armstat's. For countries that do not report (Libya, Venezuela, Syria, Iran, North Korea and others), tanker tracking and official statements fill the gap; only the share Russia was paid for counts, not supplies on credit or as aid.

### Why UN Comtrade and not Eurostat for EU members

The two agree for the EU as a whole but can differ a lot for one country (the Netherlands, 2021: EUR 18.5bn in UN Comtrade, 26.5bn in Eurostat). Eurostat books goods where they enter the EU, so Russian goods landing in Rotterdam for Germany count for the Netherlands. UN Comtrade holds each country's own figures for goods bought for its own economy, which answers "who paid Russia", with one definition for every country in the world.

## The pre-war level and the comparison

Each flow is compared with its expected no-war level, usually 2021 at pre-war prices ([counterfactual](https://warbalance.org/method/counterfactual/)). How the four flows enter the map:

| flow | part on the map | counted at |
| --- | --- | --- |
| bought from Russia | less or more paid to Russia | Russia's net gain on that mix of goods |
| sold to Russia | income lost or gained | seller's net gain × genuinely extra |
| bought from Ukraine | bought more or less from Ukraine than without the war | Ukraine's net gain; more counts for a country, less against it, against what Ukraine could export |
| sold to Ukraine | extra sales to Ukraine | seller's net gain × genuinely extra, subtracted as money flowing back |

### Purchases from Ukraine, both ways

Purchases from Ukraine count both ways, like purchases from Russia (round 60): buying more than without the war counts toward Ukraine, buying less toward Russia, each at Ukraine's net gain on what it sells (2 to 7 cents per euro), so a shortfall of EUR 10bn counts as about EUR 0.5bn.

The no-war level takes out Ukraine's own losses. Ukraine could not sell what it could not make or ship: its ports were blocked in 2022, farmland was occupied and the steel works of Mariupol were destroyed. In 2023 it exported 27% of its 2021 steel and 28% of its iron ore, but 77% of its grain and all its oilseeds. So each buyer's 2021 purchases are scaled, product group by group (grain, oilseeds, other food, iron ore, steel, other metals, wood, machinery, chemicals, other), by what Ukraine actually exported of that group each year against 2021: the supply factor. A country that kept its pre-war share of what Ukraine could still sell is neutral; one that took more helped Ukraine; one that took less (an import ban, a switch to other suppliers) counts the other way. The war's damage to Ukraine's exports is held against no one. Poland (+EUR 0.41bn), Germany and Türkiye kept buying close to their 2021 amounts while Ukraine's exports fell; China (−EUR 0.57bn) and India bought far less of what Ukraine could sell.

Bans are measured against the no-war level. Poland, Hungary and Slovakia have banned some Ukrainian farm goods (wheat, maize, rapeseed, sunflower seed) since September 2023. Before the war they bought almost none of them (Poland EUR 47m in 2021), so against a world without the war the bans change little (Poland about −EUR 0.01bn at Ukraine's gain). What they ended was the extra trade of 2022–2023, which counted for these countries in those years.

Belarus bought almost nothing from Ukraine after the invasion, because Ukraine stopped trading with it. That counts (−EUR 0.16bn): Ukraine's embargo is a foreseeable answer to Belarus hosting the invasion, as Russia's gas cut-offs are counted for the buyers of Russian gas.

How sure. The range keeps the measured band of the no-war level, applied to the lower expected level, and it is wide (China −3.4 to +0.2bn). The method band holds the other reasonable choice: the land neighbours (Poland, Hungary, Slovakia, Romania, Moldova) measured against Ukraine's exports to its neighbours and everyone else against its exports to the rest, which treats the overland rerouting of 2022 as forced. It lowers the neighbours' credit (Poland about −0.2bn instead of +0.4bn) and raises distant buyers'. The EU opened the land routes on purpose (Solidarity Lanes, duty-free access), so those purchases were partly a choice, and the central value uses all of Ukraine's exports.

## Profit, not revenue

Selling EUR 10bn less to Russia does not cost the seller EUR 10bn: the goods would have cost money to make. So a sale counts at the seller's net gain (taxes plus profit above a normal return) times the share that is genuinely extra (not sold elsewhere instead), the [same rule for every seller](https://warbalance.org/method/overview/#m-overview-one-rule-for-every-seller). Manufacturing earns little above a normal return, so sales given up count at a small share of their value. Counting gross revenue would put Cyprus at +30% of GDP and Lithuania at +45%.

## Re-export hubs

Goods made in China or the EU and sold on to Russia through Armenia or the UAE count for the country that made them, not for the hub. This applies to 11 hubs (the UAE, Armenia, Kazakhstan, Türkiye, Kyrgyzstan, Georgia, Uzbekistan, Azerbaijan, Serbia, Mongolia, Tajikistan), with measured re-export shares where they exist (Armenia 0.86–0.92, Kazakhstan 0.77–0.84). The hub keeps its trading margin (goods 5–20%, an educated guess). Russian gold and diamonds passed on by hubs go to their first buyers. Hubs keep 2021 as their no-war level.

### Belarus

Belarus's re-exports to Russia (mostly EU used cars in 2022–2023, and machinery from China, Türkiye and the UAE) are estimated from what partners shipped to Belarus, and moved to the countries that made the goods (China 45%, Germany 16%, Poland 15%).

## Baltic and Finnish pass-through

Before the war much of the Baltic states' trade with Russia was transit (Estonia's oil terminal at Muuga, Lithuanian re-exports). Counting it as "given up" would make Lithuania +45% of GDP. So the transit share is taken out before and after the war alike. It is not handed to the final buyers: they already record those goods as imports from Russia.

## Residual gap

Russia's own customs record EUR 29–46bn a year more imports than its partners report selling. Part is valuation; part is unrecorded transit through Kyrgyzstan and Kazakhstan (given to the countries of origin); the rest is spread over sellers by their reported sales. Belarus is left out: it reports its full trade with Russia.

## Hong Kong gold

Nearly all of the rise in Hong Kong's imports from Russia since before the war is precious metals (measured from its own records). For January–August 2026 the gold is measured: Hong Kong's imports of Russian-origin gold from its statistics department (C&SD Trade-IDDS); only the rest, about 7%, is estimated. The share it passed on to mainland China counts for China, the share sent elsewhere for those buyers; Hong Kong keeps the rest. Gold is valued at the 2021 gold price held flat ([prices](https://warbalance.org/method/prices/#m-prices-gold-at-its-no-war-price)). Hong Kong's direction is uncertain.

## Gold and hidden channels

Money that appears nowhere in customs: Kyrgyzstan's gold gap (partners record more gold from it than it exports; passed on to the buyers of Kyrgyz gold), and high-end allowances for China, Türkiye, Kazakhstan, Armenia and India. Certainty D. Payments through agents and crypto are not added again: they settle trade customs already record.

## Payment agents

Agents in third countries settle much of Russia's trade outside normal banking and charge fees (3–4% in 2022–2024). The share of those fees kept by local residents, about EUR 7bn over 2022–2026 (China 1.7bn, Kazakhstan 1.1, the UAE 1.0), counts as money earned from Russia, at the financial industry's net gain: about a tenth of the fees. The country split is an educated guess (D).

## Remittances and relocants

A marker, not in the totals (△). Migrants' wages sent home from Russia are individuals' income, not a country's choice; the real gain over working elsewhere is small and hard to measure; and they would dominate the bottom of the ranking (Tajikistan about −40% of GDP). The marker shows the change against the expected level since 2022: Tajikistan EUR 9.4bn more than expected (81% of GDP), Uzbekistan 18.9bn more (19%). Most of the rise comes from Russia's war economy pushing wages up.

Relocants are left out: money Russians brought when they moved abroad after 2022 is money leaving Russia, about EUR 30bn of the 2022 surge in transfers (Armenia, Uzbekistan, Georgia...). Their living spending is also taken out of services sold to Russia.

## Narrower actual values

National and partner statistics narrowed ten series in nine countries: Armenia's gold passed on, Brunei's refinery, Tajikistan's and Belarus's official totals and prices, the land neighbours' own-report margins, and Yemen's and Syria's pre-war levels.

## Goods Russia's weapons need: the critical-goods marker

A euro of microchips matters more to Russia's war than a euro of bananas, but the map counts money. So sales of the 50 goods on the allies' Common High Priority List (chips, electronics, bearings, machine tools) are shown as a marker (◆), not added: the money is already in the trade number. About EUR 42bn over 2022–August 2026: China 23.6bn, Hong Kong 7.5, the UAE 1.8, India and Türkiye 1.4 each. EU members sold EUR 3.8bn in 2021 and almost nothing from 2023. Large sellers get the tag "Sold Russia goods its weapons need" ([tags](https://warbalance.org/method/tags/)).

## Freight: CIF to FOB

Import values include freight and insurance, which go to shippers and insurers, not Russia. Each importer's own freight margin (OECD data: China 6.4%, India 7.2%, Germany 4.9%) is taken out of its payments to Russia, except for trade priced at the Russian border (pipelines, overland).

## January–August 2026

2026 trade comes, for 32 economies, from monthly UN Comtrade data and national statistics, not from the 2025 rate (India's imports from Russia: EUR 48bn, not 38bn). Aid is the Kiel Institute's release of 8 October 2026, data to 31 August 2026.

## Smaller corrections

Denmark's sales to Russia sit between Danish and Eurostat figures. Small-country data artefacts are fixed (Lebanon's 2019 fuel for smuggling into Syria, traders' origin labels in Mauritania, storage terminals in Saint Lucia and Bonaire). Dual-use goods are inside recorded sales and not added again.

## Sanctions-mandated or own choice

A label only; totals do not change. For 42 economies with sanctions, it estimates which share of each cut was required by bans (EU Regulation 833/2014 and national regimes). A ban counts as "required" only for what was still flowing when it took effect: a fuel a country stopped before the ban was its own choice. Required share of the fall, 2022–2026: Lithuania 21%, Finland 27%, Poland 38%, Germany 30%. Poland's pipeline crude, which Russia halted in February 2023, is labelled as cut off by Russia, like the gas Gazprom cut off. For a gas cut after 2022 (Austria 2024 by Gazprom; the end of transit through Ukraine in 2025, which was Ukraine's decision not to renew its contract), only the gas still flowing the year before is labelled as stopped: about 4% of Italy's fall, 30% of Austria's.

## The sanctions premium

Sanctions make Russia pay more for what it still buys: the same goods cost Russia more than the same seller charges others. Until round 38 the map read that higher price as more volume, counted at the thin margin of an ordinary sale. Since round 39 it is measured: each seller's unit values to Russia against the same goods to everyone else, against 2021 (UN Comtrade, by product, fuel and gold left out). The premium comes off the seller's sales to Russia, and the part that really stays with the seller (not compliance and payment costs, changes in the mix, over-invoicing or Russian-owned front companies; about half for producers, a quarter for trade hubs) is a part toward Russia. About EUR 23bn (5 to 74) in all, mostly China (+0.1% of GDP toward Russia), then Türkiye, Kazakhstan and Uzbekistan. Re-exports keep their hub margin instead, so nothing is counted twice.

## Russia's gain and the country's own gain

"Russia's earnings from trade" adds two different things, so the data split it in two lines per country and year, each with its own range: Russia's own gain, what Russia itself kept more or less, and the country's own gain from Russia, what the country's firms and state earned from selling to Russia. Both stay in the total: a country that earns more from Russia's war economy counts toward Russia, as a seller to Ukraine counts toward Ukraine's side. The two lines add up to Russia's earnings from trade; their ranges do not add up.

| part | line |
| --- | --- |
| Less or more paid to Russia (fuel and goods Russia sold) | Russia's gain |
| War-time fuel prices (the windfall, after any discount) | Russia's gain |
| Russian oil via other countries | Russia's gain |
| Services bought from Russia | Russia's gain |
| Taxes paid in Russia by the country's companies | Russia's gain |
| Profits sent home by Russian-owned firms in the country | Russia's gain |
| Weapons bought from Russia; Wagner and the Africa Corps | Russia's gain |
| Rosatom and Russian state loans | Russia's gain |
| Gold and hidden payments to Russia | Russia's gain |
| Weapons and troops given to Russia (the unpaid share) | Russia's gain |
| Sales to Russia, and the sanctions premium on them | country's own gain |
| Services sold to Russia (tourism, business services) | country's own gain |
| Shipping and insuring Russian oil, tankers sold to the shadow fleet | country's own gain |
| Payment agents' fees | country's own gain |
| Its companies' profits in Russia and stakes in Russian oil and gas projects | country's own gain |
| Weapons sold to Russia (the paid share) | country's own gain |

Examples, 2022 to August 2026: China raised Russia's own gain by EUR 102.6bn and earned EUR 25.3bn more from Russia itself; Germany cut Russia's own gain by EUR 64.6bn and gave up EUR 8.0bn of its own earnings; most of Greece's figure is its own gain (EUR 1.26bn toward Russia, mostly tankers sold to the shadow fleet), while Russia's own gain from Greece moved by EUR 0.29bn. In data.json the lines are rusGain and ownGain (years: each year's change; cum: the total to each year with its range, EUR bn).

[How it works](https://warbalance.org/how-it-works/) · [Limitations](https://warbalance.org/limitations/) · [Changelog](https://warbalance.org/changelog/) · [Open in the map](https://warbalance.org/#view=how)

Cite as: War Balance, figures to August 2026, https://warbalance.org/method/trade/