How this is measured, and where it can mislead

What these maps are

Atlas of Influence draws one measurable thing per map. Plate 01 and the stories use trade: how much of each country's goods trade is with a given power. That is a measure of commercial dependence, not of loyalty or control. A country can trade heavily with a power it distrusts. The maps are a starting point for a question, not a verdict.

Source

IMF International Merchandise Trade Statistics (IMTS), the successor of Direction of Trade Statistics: https://data.imf.org/en/datasets/IMF.STA:IMTS. Indicators used: exports of goods, FOB, and imports of goods, CIF, annual, in US dollars, by partner country, 1995–2025. Processed 2026-10-07. The latest year can be revised by the reporting countries later.

Definitions

• Trade between A and B = A's exports to B plus A's imports from B, as reported by A. If A did not report a flow, B's report of the same flow is used (a "mirror" figure). If one year is missing between two present years, the average of its neighbours is used.
• A country's total trade = its own reported world total. If it reported none, or if the sum over partners is more than 5% above its own total, the sum over partners is used and the tooltip says "built from partner reports" (593 country-years in all).
• Share with a power = trade with the power's member countries ÷ the country's total trade.
• Largest partner = the single biggest partner when each tracked power counts as one block (the EU counts as one) and every other country counts alone.
• Reach (the story charts) = the share of countries with data that do at least 10% of their goods trade with the power. Every country counts once, so Luxembourg weighs the same as India. The power's own members are left out.

Years that are withheld

Some country-years have a flow that is missing in both directions even though neighbouring years have it. If such gaps add up to more than 4% of the country's total, that country-year is drawn hatched rather than shown with a silently understated total. The table at the bottom lists how many countries are withheld each year.

Choices that change the picture

• The EU is today's 27 members in every year, including years before some of them joined and including the years the UK was a member. It is a fixed block, not the union as it was at the time.
• China is mainland China. Hong Kong and Macao are separate partners. Taiwan does not report to the IMF, so every Taiwan figure is built from partner reports.
• Borders are Natural Earth 1:10m, in its version drawn from Ukraine's point of view, so Crimea and the occupied parts of Donetsk and Luhansk are Ukraine, as in UN recognition. Trade statistics follow each reporting authority and do not separate those areas.
• Powers are listed in a data file. Adding one is a data change, not a redesign.

Where trade data misleads

• It counts goods only, not services, finance, people or security.
• Hubs distort the partner. Goods bound for a country can be recorded as trade with the port or re-export hub they pass through, such as the Netherlands, the UAE, Singapore or Hong Kong.
• Imports are valued with freight and insurance (CIF), exports without (FOB), so the two sides of one flow do not match exactly.
• Sanctions and secrecy leave gaps. Where a country's own report is missing, partner reports fill in, which is why a sanctioned economy can still be drawn but is probably understated.
• Counts of countries are unweighted and the number of countries with data changes from year to year (table below), so charts use shares of countries with data rather than raw counts.
• Countries that split or merged (Sudan and South Sudan, Serbia, Montenegro and Kosovo) are drawn with today's borders, but their statistics may be reported on a different basis in earlier years.

Data

The data behind every map is one file: /data/trade.json. For each country and year it holds total trade, a quality flag (0 own total, 1 built from partners, 2 withheld), the largest partner, and the share with each tracked power, in tenths of a percent.