Europe buys raw material. And sells machines back.
82 percent of what the EU imports from sub-Saharan Africa is primary goods. 67 percent of what it sells back is manufactures. The numbers come from the European Commission's own statistics — and three of the things I set out to prove did not survive checking.
I set out to establish seven things in this piece. I established four. On three, I could not find a single number that survives an adversarial check. What those three are is at the end — because their absence is itself a finding.
Start with what holds.
The structure
The European Commission publishes a factsheet on goods trade with the African ACP states. It is public, it is their own statistic, and it describes the relationship more precisely than any brochure would.
- EU imports from the region
- EU exports to the region
Datentabelle
| EU imports from the region | EU exports to the region | |
|---|---|---|
| Primary products | 74.935 EUR m | 22.552 EUR m |
| Manufactures | 13.874 EUR m | 48.125 EUR m |
Values in EUR million. Shares do not sum to 100 percent — “other products” account for 2.1 percent of imports and 0.4 percent of exports.
Of the EUR 91.4 billion the EU imported from the region in 2025, 82.0 percent were primary products and 15.2 percent manufactures. Of the EUR 71.6 billion it sold back, 67.2 percent were manufactures. The largest import section is mineral products (37.2 percent); the largest export section is machinery and appliances (25.4 percent).
This is not an outlier year. The 2022–2025 series shows the same shape, year after year.
Raw for processed. That is not an accusation. It is a measurement.
Three things have to follow that chart immediately, or it lies.
First, the geography. "African ACP" means 48 states south of the Sahara, South Africa included. North Africa is not in it. Total EU–Africa goods trade ran to roughly EUR 356 billion in 2024, so these figures cover well under half of it. Label this chart "EU–Africa" and you understate the relationship by more than a factor of two.
Second, the definition. The Commission's "primary products" grouping includes SITC code 68: non-ferrous metals. Smelted, refined metal counts as primary. Read "unprocessed" out of this chart and you have read in more than it says.
Third: shares that do not sum to a hundred are not an arithmetic error. They are the residual category.
The capital runs one way
- EU stock in Africa
- Africa’s stock in the EU
Datentabelle
| EU stock in Africa | Africa’s stock in the EU | |
|---|---|---|
| 2021 | 267.535 EUR m | 117.677 EUR m |
| 2022 | 279.370 EUR m | 108.042 EUR m |
| 2023 | 242.799 EUR m | 68.060 EUR m |
| 2024 | 250.668 EUR m | 74.639 EUR m |
Values in EUR million. Net positions on an immediate-counterpart basis, special-purpose entities included — on a gross basis the 2023 EU stock is some EUR 19 billion higher.
Quelle Eurostat, dataset bop_fdi6_pos (BPM6), net positions, immediate counterpart, data vintage 09-06-2026
In 2024 the EU held EUR 250.7 billion of direct investment stock in Africa. Africa held EUR 74.6 billion in the EU. Ratio: 3.4 to 1.
One warning here too, and this one cuts the other way: the "Africa's stock in the EU" series has a structural break in 2018/19 — it jumped 265 percent in a single year. You therefore cannot build a story about African capital behaviour out of its movements. I am not building one.
Where that capital supposedly sits — and where it actually does
This is where it gets interesting, because the obvious headline falls apart on inspection.
Book the EU's African stock by sector and finance dominates: 38.9 percent, against 17.6 percent in manufacturing. From which one could build a nice, sharp thesis: Europe does not invest in production in Africa, it invests in financial vehicles.
Except it isn't true in that form.
- as booked
- excluding shell entities
Datentabelle
| as booked | excluding shell entities | |
|---|---|---|
| Financial and insurance activities | 38,9 % | 23 % |
| Manufacturing | 17,6 % | 22,6 % |
| Mining and quarrying | 9,8 % | 12,2 % |
Share of the respective total stock, in percent. Special-purpose entities account for 22.7 percent of the total stock — and 54.3 percent of the “financial services” line alone. Eurostat's own metadata and the ECB both recommend excluding them for economic analysis.
Quelle Eurostat, bop_fdi6_pos (BPM6), 2023 positions, EU27 → Africa, by NACE Rev. 2; entity = TOTAL versus entity = SPE
More than half of the finance line is conduit vehicles — pass-through structures, not money working in Africa. Strip them out and finance (23.0 percent) and manufacturing (22.6 percent) are effectively tied.
The sharp thesis dies in the footnote. I wrote it down anyway, because the dying is the point: the booked number and the economic number are two different things, and anyone quoting only the first is arguing from an artifact.
The price of sending money home
Datentabelle
| Cost | |
|---|---|
| Middle East & North Africa | 5,1 % |
| South Asia | 5,3 % |
| Latin America & Caribbean | 5,6 % |
| East Asia & Pacific | 5,8 % |
| Europe & Central Asia | 6,8 % |
| Sub-Saharan Africa | 8,5 % |
Average total cost as a percent of the amount sent, across all sampled sending countries. This is explicitly NOT a Europe→Africa figure: the Sub-Saharan average includes the expensive intra-African corridors. UN Sustainable Development Goal 10.c sets the target at 3 percent.
Quelle World Bank, Remittance Prices Worldwide, Issue 54 (Q3 2025)
8.46 percent. Sub-Saharan Africa is the most expensive region on earth to send money to — nearly three times the UN target of 3 percent, and well above the global average of 6.36 percent. Nine of the world's thirteen corridors costing more than 20 percent originate in Sub-Saharan Africa.
The World Bank names the causes: too little competition among providers, and missing cross-border interoperability. Those are not laws of nature. They are design faults, and design faults can be fixed.
A chart I am not drawing
There is a claim in circulation that China has overtaken the EU as Africa's largest trading partner. It can be proven — and disproven. With the same document.
On page 8 of the Commission's factsheet, computed on IMF data: China EUR 157.4 billion, EU27 EUR 139.9 billion. China ahead.
On page 2 of the same PDF, computed on Eurostat Comext: EU27 EUR 163.0 billion. That is EUR 23.1 billion, or 16.5 percent, higher — and therefore ahead of China.
Same authority, same document, same year, two datasets, opposite results. For a like-for-like comparison of two partners the IMF series is the methodologically correct one, so: China is ahead. But put the two series in one chart and you have produced a lie out of correct numbers.
So there is no chart here. A bar chart would have hidden the conflict. A paragraph can show it.
What the relationship is worth, to whom
The African ACP states accounted for 3.2 percent of EU external trade in 2025. The EU accounted for 21.4 percent of that region's trade.
And even the 3.2 percent flatters Europe: the denominator is extra-EU trade. Count trade between member states as well — which is the economically correct thing to do — and the share falls to roughly 1.5 percent. The asymmetry is larger than the first number suggests, not smaller.
Alongside it, a trend nobody in Brussels enjoys presenting: the EU's trade balance with the region swung from +EUR 5.0 billion (2015) to −EUR 19.9 billion (2025). EU exports to the region have contracted three years running. In nominal terms they grew 2.3 percent across the whole decade — in real terms, a marked decline.
In fairness: the swing is substantially an energy-price effect — mineral fuels alone are 32.4 percent of EU imports from the region. And for Africa as a whole, North Africa included, the picture is much milder. This finding belongs to the ACP grouping, not to the continent.
The three holes
I also set out to establish: what energy and critical raw materials Europe draws from Africa and where those materials are processed; where the value accrues in subsea cables and data centres; and how much of the EUR 150 billion the EU announced for Africa under "Global Gateway" has actually been disbursed.
On none of those three did I find a number that survives checking.
That is remarkable — not because the data does not exist, but because what circulates freely disintegrates the moment you press on it. A widely cited figure on the collapse in project finance failed. The cocoa-versus-chocolate example, the standard illustration of European tariff escalation, failed. The claim that remittances exceed foreign direct investment by half again, failed. The claim that Europe is Africa's largest investor, failed.
The Global Gateway disbursement rate in particular is probably the single most load-bearing number in this whole series — and the worst evidenced. That is not a footnote. If the central metric of a EUR 150 billion promise is not publicly verifiable, that is the story.
And the tariff escalation I had counted on? The mechanism is real and documented by UNCTAD: processed goods are taxed more heavily than the raw material they are made from. But for signatories of the Economic Partnership Agreements, and for the least-developed countries, the EU charges zero duty — on both. Escalation bites mainly on those who are neither. Nigeria is the obvious test case. And the binding constraint there is not the tariff schedule at all: it is rules of origin, and the cost of proving compliance with sanitary and phytosanitary rules.
That too is a result. The story I expected was the wrong one.
What is left
The entanglement is real, it is measured, and it is asymmetric in every dimension that holds up. Europe buys raw material and sells machines back. European capital in Africa is three and a half times African capital in Europe. Sending money home costs more here than anywhere else on earth.
At the same time, Europe is losing the relationship it claims to care about: exports have been shrinking for three years, and on trade volume, China has gone past.
That is the structure. Whether it is good is a separate question, and it has more than one answer. The next two parts go looking for it.
Every number in this piece comes from a primary source and carries that source under the chart it appears in. Claims that did not survive checking are not in it — not even the ones that would have supported my argument.