We read all 68 reporters: the low-income trade-tax median (15.3%) is 16x the high-income median (0.95%); US 46th of 68

A trade correspondent covers one tariff at a time — who raised which duty on whom. We read every government that reported the 2024 series and sorted them on one line, and the shape only the full read shows is this: income orders the panel at the median, but structure sets the top, and the largest economies sit low. Here is the record behind that. The World Bank compiles, from the IMF's Government Finance Statistics, an indicator called "Taxes on international trade (% of revenue)" (series GC.TAX.INTT.RV.ZS). The World Bank defines it exactly: "Taxes on international trade are taxes that become payable when goods cross the national or customs frontiers of the economic territory or when transactions in services exchange between residents and non-residents. This indicator is expressed as a percentage of revenue which includes all transactions that add to the amount of economic value of a unit or sector." The numerator is therefore wider than import and export duties alone: it also covers taxes on services exchanged between residents and non-residents. The denominator is revenue as the definition states — all transactions that add economic value — not tax revenue only. The series sits in the World Bank's GC (central-government) family. For the 2024 observation year in the current database vintage (last updated 2026-07-13), a query filtered to `date=2024` returns a non-null value for 68 reporting governments once the 11 regional and income-group aggregates are removed. We list the 68 entity codes in the methodology below, so the panel is auditable. This is a panel of the governments that reported the series for 2024, not the world: about 217 economies exist, and most filed no 2024 value. Ranked highest to lowest, the panel opens: Namibia 32.7%, the Bahamas 23.6%, the Philippines 20.8%, Somalia and Ethiopia at 18.5%, Tanzania 16.9%, Burkina Faso 16.6%, Guinea-Bissau 16.0%, Argentina 15.0% and Madagascar 14.6%. Namibia's reading sits 9.1 points clear of the second-placed Bahamas — the widest gap anywhere in the panel. Among the 68 reporters, three drew at least 20% of revenue from trade taxes, eight at least 15%, fifteen at least 10% and twenty-five at least 5%. The median across the 68 reporters was 3.6%. Within the panel, the income-group medians fall in order. The eight low-income reporters posted a median of 15.3% — Somalia and Ethiopia both 18.5%, and only Mozambique (5.3%) below 6%. The 11 lower-middle-income reporters posted a median of 8.5%, the 25 upper-middle-income reporters 3.1%, and the 24 high-income reporters 0.95%. The low-income median is 16.0 times the high-income median. Only three high-income reporters cleared 5%: the Bahamas (23.6%), Andorra (10.1%) and Nauru (5.3%). At the bottom of the panel sit high-income and customs-integrated reporters: Greece 0.01%, the United Arab Emirates 0.01%, France 0.003%, Romania 0.0003% and Bosnia and Herzegovina 0.00%, all rounding to 0.0%. The United States reported 1.60% — 46th of the 68, below Brazil (2.5%), Mexico (2.2%) and South Africa (3.6%), and above China (1.3%). The United States (1.60%), China (1.25%), Brazil (2.49%) and Mexico (2.15%) each reported under 2.5%.

The following is analysis, not fact. The ratio measures one thing: of every dollar of central-government revenue in 2024, how many cents came from taxes on international trade. Read across the 68 reporters at once — which no single-country desk does — the panel sorts by income at the median, and several of the largest economies in it sit in the lower half: the United States (46th), China (50th), Brazil (42nd) and Mexico (43rd). The United States, at 1.60%, draws a smaller share of its revenue from trade taxes than 45 of the 68 reporters, and a fraction of the 15.3% median across the eight low-income reporters. For the United States the trade-tax share of central revenue is small, at 1.60%; for Burkina Faso (16.6%) or Somalia (18.5%) it is a large share of reported central revenue. This is a share, not a measure of budget dependence: the panel carries no spending, substitution or enforcement-cost data, so it cannot show how load-bearing the tax is ^mmodeled. The income ordering holds at the median, but income does not set the top of the panel. The highest reading belongs to Namibia, a lower-middle-income economy, and the second to the Bahamas, a high-income one. What the two share is not income but trade structure ^mmodeled. Namibia and Lesotho (13.0%, 13th) are both members of the Southern African Customs Union, whose public revenue-sharing pool collects the bloc's customs and excise receipts and redistributes them, with a formula weighted toward the smaller members. Whether that pool explains their readings depends on how the IMF classifies redistributed SACU revenue in each country's numerator and denominator, which this series does not disclose; we present it as a hypothesis, not a finding ^mmodeled. The Bahamas, Andorra and Nauru are the only high-income reporters above 5%; a common feature is a small, open economy with a narrow domestic tax base, but the series does not establish that as the cause ^mmodeled. The bottom of the panel may reflect accounting as much as policy ^mmodeled. Customs duties collected at the external frontier of the European Union are treated as an "own resource" of the EU budget, not as national revenue. A member state's central-government accounts could therefore record little trade-tax revenue even as goods are dutied at its ports ^mmodeled. France's 0.003% and Romania's 0.0003% are consistent with that treatment, but we have not confirmed how each country's GFS submission books EU own resources, so we cannot state it as the reason ^mmodeled.

The most important caveat is coverage, and it cuts against any claim of a complete ranking. We read all 68 reporters, but 68 is not the world: only 68 of roughly 217 economies reported this series for 2024, and many economies are missing. We do not know why any given economy filed no value, and we do not claim to — non-reporting can follow from reporting capacity, timing or classification, and this series does not say which ^mmodeled. "46th of 68" is 46th among reporters, not on Earth. Because the missing economies are absent rather than measured, coverage prevents any inference about the shape or width of the global distribution — the unobserved values could sit high or low ^mmodeled. A second limit is the accounting boundary, and it may matter most at the bottom. If an EU member's near-zero reading reflects where duties are booked rather than whether the border is taxed, then comparing an EU member's 0.0% with a non-member's 15% is not like-for-like ^mmodeled. The denominator adds a further caution: resource-rich states booking large non-tax receipts, and customs-union members receiving pooled transfers, can sit high or low for reasons unrelated to their own tariff schedules ^mmodeled. Third, the ratio is a single-year 2024 snapshot; it ranks a revenue share, and it does not measure a tariff rate. A government can raise tariffs and still show a low share if its other revenue is large. Finally, timing. This is a 2024 observation. It predates the tariff changes several large economies enacted or debated afterward, and it cannot measure revenue from any policy that took effect after 2024 ^mmodeled. A low 2024 share is not evidence about the size of a later tariff, and this piece makes no claim about post-2024 policy.

From a finance ministry among the low-income reporters, a share near a sixth of central revenue is a large line in the accounts — and, if trade slows or a customs-union formula is renegotiated, a large line to defend ^mmodeled. From a treasury among the high-income reporters, the same 1.6% is a small share of central revenue. The identical series carries very different weight in the two budgets, though the panel alone cannot say how substitutable that revenue is on either side ^mmodeled. We do not sit in either ministry; what we can do that neither can is read both budgets on one line at once. Laid there, the series shows the gap in the share itself: 15.3% at the low-income median against 0.95% at the high-income median.