War Balance

Aid, refugees, military support and flowback

The Ukraine side of the ledger: government aid, each EU member's share of EU-level aid, undisclosed military aid, private donations and the cost of hosting refugees. On the Russia side, weapons and troops sent to Russia. In between, flowback: aid that returns to the donor's own economy as orders for its own industry.

Details for developers: lead times, factors, sources and research files.

Aid

Aid comes from the Kiel Institute Ukraine Support Tracker (release of 8 October 2026, data to 31 August 2026): allocations (delivered or contracted), not pledges. For the 190 countries outside Kiel, UN and government records fill the gap; countries with no record anywhere get zero with a small upper bound.

Four adjustments, which is why the map's figure differs from Kiel's (the panel shows both):

From Kiel to the map, € billion, bilateral aid to August 2026 (each member's share of EU-level aid is a part of its own). "Other": countries outside Kiel and rounding.

Donor Kiel G7 loans Russia repays Not yet delivered Loans at gift value Weapons at market value Other Map
United States 115.38 −18.34 −10.20 0.00 −7.01 +0.01 79.84
Germany 30.84 0.00 −4.33 −0.16 −0.49 0.00 25.86
United Kingdom 22.80 −2.65 −1.86 −3.54 −2.67 +0.01 12.09
Canada 14.85 −3.26 −1.31 −3.48 −0.18 0.00 6.62
Japan 14.42 −1.58 −0.01 −7.94 −0.03 +0.01 4.87
Denmark 11.74 0.00 −1.98 −0.03 −0.68 +0.01 9.06
Norway 11.53 0.00 −2.04 0.00 −1.06 +0.01 8.44
Sweden 11.33 0.00 −2.04 −0.26 −2.44 0.00 6.59
Netherlands 11.20 0.00 −0.97 −0.22 −1.31 −0.01 8.69
France 7.91 0.00 −1.48 −0.57 −1.59 −0.01 4.26
Poland 5.89 0.00 −0.28 −0.87 −1.78 −0.01 2.95
Italy 4.31 0.00 0.00 −0.18 −0.75 0.00 3.38
All countries 283.12 −25.83 −28.11 −17.93 −24.68 +0.65 187.22

Refugee hosting is not in Kiel's aid. For EU hosts, 2022–2025, Kiel's modelled cost (cited by the European Parliament's research service) is €163bn; the map counts €74bn: what governments actually spent (€79bn), minus what donors paid back and the host's gain on refugees' work. Kiel models about €10,800 per refugee-year; the budgets paid about €5,300. Spending per refugee differs widely: the Netherlands about €24,000 a year, Poland about €2,400.

EU-level aid

Aid from EU institutions is split over the 27 members by their share of the EU budget, Kiel's own method. The 2026 EU Support Loan is guaranteed by 24 members, so Czechia's, Hungary's and Slovakia's slices go to the other 24. The European Peace Facility is not counted: it reimburses members for weapons already in their own aid, so adding it would count them twice (limitations).

Donated weapons at market value

Donors value weapons from their stocks very differently (the US at replacement cost until 2023, the UK at replacement cost far above book value). The map counts what the equipment would fetch on the market: old MiG-29s at a fraction of their recorded value, shells above it because their price doubled. This lowers bilateral military aid by about 14% (EUR 177bn → 152bn), most for the US, the UK and Sweden; Germany's mostly new purchases barely move.

Loans at their economic value

A loan that is repaid is worth less than a grant. Each loan counts at its gift: cheaper interest plus the default risk the lender really carries. Over 2022–2026, EUR 96.5bn of loans at face value become about 55bn. The full method is on the loans page. New IMF, World Bank and EBRD lending counts only above pre-war levels, credited by shareholding.

Undisclosed aid and private donations

Refugees

Hosting refugees from Ukraine counts toward Ukraine at what all levels of government actually spend on them, minus what donors paid, minus the host's real gain on refugees' work (the extra profit and taxes on production their work creates). Refugees' own wages and the taxes on them are their income, not the host's gain, so they are not subtracted.

Refugee spending flowing back

Money spent hosting refugees is spent in the host's own economy (rent, food, shops). Only the host's real gain on it is subtracted, under the one rule: about 2.7% of the hosting cost overall (Germany 2.2%, Poland 7.3%).

Military support to Russia

"Weapons and troops sent to Russia", always toward Russia:

Paid deliveries count as sales: what Russia paid for, also in kind (technology, fuel, food), counts at the seller's net gain, like any sale; what it did not pay for counts in full. North Korea was largely paid in kind; about half its shells were old stock with no other use. A method-band alternative counts only what was seen flowing back to Pyongyang, which moves North Korea's net from −19.9% to −32.6% of GDP. North Korea is a party to the war and not ranked.

The seller's gain on these sales is worked out from price against cost, item by item (round 60), not from Western arms makers' average (5 cents per euro), because a state monopoly selling to one buyer with few alternatives keeps far more:

These sales exist only because of the war: no other buyer was turned away for them, so little of the gain is displaced business, and 0.9 (0.8–1.0) of it counts as genuinely extra, not the 0.6 used where a war sale can crowd out other buyers (the round-60 choice). Iran's weapons now count for EUR 1.84bn of the 3.5bn delivered (before: 0.11bn), North Korea's for EUR 5.87bn (before: 3.33bn). Belarus (not paid) and China's unrecorded part (sold by private firms) are unchanged.

The Russia panel's list "Armed Russia" shows all deliveries at full value (North Korea EUR 8.8bn, Iran 3.5bn, China 1.3bn, Belarus 1.3bn); the totals do not change.

Sources and values

supplier 2022–Aug 2026, EUR bn (range) certainty main evidence
North Korea 8.8 (3.3–17.6) C 6.5–15m shells, 150–280 missiles, troops' pay
Iran 3.5 (1.5–5.5) C drones 2022–23, the Alabuga licence, ballistic missiles
Belarus 1.3 (0.55–2.7) D ammunition from army stocks
China 1.3 (1.0–3.0) D unrecorded military items

Volunteers for Ukraine and fighters for Russia are shown as counts, not money.

Flowback

Part of aid comes back to the donor as orders for its own industry. Counting the full aid while ignoring this would flatter big arms producers. Each seller's flowback is its own aid spent at home (including replacing donated stock), other countries' aid spent there, Ukraine's own arms purchases there, aid spent there on energy and rebuilding, and extra sales to Ukraine. It counts at the seller's net gain × genuinely extra: arms makers earn little above a normal return, so the US earns back only a few percent of the money spent in its economy. The "money flowing back" switch on the map turns it off.

Services sold to Ukraine

Since round 39 extra services sold to Ukraine count as flowing back too, like extra goods: transport on the solidarity lanes and other services the sellers report (Eurostat, OECD-WTO), and Ukrainians' own card spending abroad (the National Bank of Ukraine), spread over hosts by refugee numbers. Each at the seller's real gain times the share that is genuinely extra; host governments' refugee spending is not in it (that is refugee hosting). About EUR 1bn in all; Poland −0.06% of GDP, the most.

Arms Ukraine bought abroad

Weapons Ukraine bought from foreign sellers, with its own budget or donors' cash (the Czech ammunition initiative, Bulgarian and Serbian shells, Patriots from the US), are money flowing back to the seller too: about EUR 21bn over 2022–2026, Czechia 5.7bn gross, Bulgaria 5.0, the US 4.1.

The all-money ledger

Flowback comes from one ledger of all money that reached Ukraine, EUR 437bn: every Kiel allocation, undisclosed aid, IMF and World Bank money, private donations and Ukraine's own arms purchases abroad, each assigned once to where it was spent. EUR 220bn was spent in Ukraine (mostly pensions, social payments and salaries), 71bn in the payer's own economy, 69bn elsewhere. Weapons given from stocks (EUR 72bn) earn nothing unless the donor bought replacements. The US earns back EUR 2.08bn, and its support for Ukraine is EUR 82.6bn (0.32% of GDP).

How it works · Limitations · Changelog · Open in the map

Cite as: War Balance, figures to August 2026, https://warbalance.org/method/aid-refugees-military/