The OECD's policy paper on Latin America's absorption of global steel excess capacity, released on 17 November 2026, has been reported mainly as an import story: combined direct and indirect steel imports into the region's six largest steel-producing economies rose 47% between 2019 and 2025, from 37.6 Mt to 55.4 Mt, with China the dominant source and roughly 40% of the total arriving as steel embedded in machinery, vehicles and fabricated goods.
That framing is accurate but incomplete. Buried in the same release is a figure with far more strategic weight for the region: trade among those six economies fell 45% over the same period. A market that is importing more is, at the same time, buying dramatically less from itself.
The arithmetic that matters
Start with the scale. At 55.4 Mt, combined direct and indirect imports into the six economies are, in the words of the OECD Steel Unit's Anthony de Carvalho at the Alacero Summit in Mexico City a week before publication, "virtually equivalent to the region's entire crude steel production" — Alacero puts regional crude output at roughly 55.7 Mt in 2025. Read that against demand of about 76.7 Mt and capacity utilisation near 60% and the picture is not one of a market being flooded in the abstract, but of an industrial base operating at little more than half its capability while an amount of steel equivalent to its entire smelting sector's output crosses its borders.
The reported metrics, collected:
| Indicator | 2019 | 2025 | Change |
|---|---|---|---|
| Direct + indirect imports, six economies | 37.6 Mt | 55.4 Mt | +47% |
| Import penetration | 33.8% | 40.3% | +6.5 pp |
| Direct Chinese imports as share of regional demand | ~7% | ~13% | ~+6 pp |
| Intra-regional trade, six economies | — | — | −45% |
| Brazil steel exports | ~12 Mt | 7.8 Mt | ~−34% |
| Mexico steel exports | ~5 Mt | 3.3 Mt | ~−34% |
| Capacity utilisation | — | ~60% | — |
Source: SteelOrbis reporting of the OECD paper. The six economies are not named in that report; context points to Brazil, Mexico, Argentina, Colombia, Chile and Peru, but that list is an inference.
The penetration figure is a trap
One numbers problem deserves flagging because it will be misquoted in trade-policy argument for the next year. A 40.3% penetration rate applied to 76.7 Mt of demand implies about 31 Mt of imports. But combined direct and indirect imports are 55.4 Mt — 72% of demand. The two cannot be on the same basis. The 40.3% figure is evidently a direct-steel-import measure, and it lines up closely with Alacero's own estimate of record import penetration at about 40–41% of regional consumption in 2025 on roughly 30 Mt of imports.
This is not pedantry. The paper's central methodological claim is that conventional customs data understate exposure because they ignore steel entering inside downstream products. The 40.3% headline is the conventional metric, and it understates the true figure by more than 30 percentage points. The OECD's separate Economic Security in Global Production work, drawing on the same analysis, puts Chinese steel at about 32% of all steel embodied in Latin America's final demand in 2024, up from roughly 17% in 2015, with domestic supply falling from about 51% to 40%. That, not 40.3%, is closer to the real exposure measure.
A second inconsistency is worth noting. de Carvalho put China at 39% of the 55.4 Mt import total; SteelOrbis reports that China and ASEAN combined accounted for "more than half" of foreign-sourced steel consumed in 2024. Different denominators, different years — consistent only if ASEAN adds a low-teens contribution. For practical purposes, the 39% China share of the 2025 import aggregate is the more precise number.
Direct, not triangulated — which changes the policy answer
The finding with the most immediate commercial consequence is the OECD's conclusion that Chinese steel reaching Latin America via third countries accounts for a relatively limited share, so direct trade remains the principal transmission channel. This cuts against the reflex explanation of trade diversion that dominates European and North American steel debates, where transshipment and downstream circumvention are central concerns.
If the flow is direct, then anti-dumping and safeguard actions aimed at named origins can, in principle, bite. The corollary is that the region's trade-defence architecture — fragmented across six national systems with limited investigative capacity, divergent deadlines and little mutual recognition — is poorly configured for a problem that is large, direct and arriving quickly. That is precisely the question the OECD raises: whether Latin America should build a regional trade-defence framework rather than continue with country-by-country cases that each take a year or more to conclude.
The integration story is the underreported one
The 45% collapse in intra-regional steel trade, and the sectors of it the OECD chose to highlight, are more diagnostic than the Chinese volumes. Chile's share of imports sourced from other major Latin American economies fell from 18% to 5% between 2019 and 2025; Colombia's from 40% to 9%. These are not markets that were integrated and got a little less so. They are markets that have fundamentally reoriented their supply chains toward extra-regional sources — largely Asia — in six years.
Meanwhile Brazil and Mexico, whose producers were the natural suppliers for those Andean buyers, saw their own exports fall by roughly a third. Both are running their own substantial domestic demand problems, but the sequencing matters: the region's largest and most efficient mills lost their export markets at approximately the same time as they lost pricing power at home.
Set against the 60% capacity utilisation figure, the strategic contradiction is stark. The region has idle steelmaking capacity of roughly 35–40 Mt a year and imports 55 Mt. It is not a shortage of capability; it is a failure of cost competitiveness and of regional logistics and commercial integration. The OECD names the causes explicitly: high energy and logistics costs, infrastructure constraints and insufficient investment. None of those are solved by anti-dumping duties, and some are arguably worsened by the input-cost effects of protection. This is where the deindustrialisation warning in the paper acquires real force — not because imports are large, but because the region's answer to them may entrench the very cost disadvantages that drive the imports.
What to watch
The paper was launched into an active policy process rather than a vacuum. The GFSEC's Milwaukee Framework for Joint Action, which Mexico, Brazil and Argentina participate in, moves toward stronger import monitoring, country-of-melt-and-pour requirements and scrutiny of suspicious trade patterns — and de Carvalho has signalled coordinated implementation beginning. Country-of-melt disclosure matters especially here: it is the mechanism by which direct flows can be traced and by which the "embedded steel" problem becomes visible to customs authorities who currently see only a machine tool or a vehicle.
For steel producers and traders, three practical implications stand out. First, expect Latin American trade-defence activity to increase in frequency and to become more coordinated, with a focus on country-of-melt data. Second, expect downstream manufacturers — the importers of that 40% indirect volume — to become political actors in their own right, because duties on steel inputs raise their costs and they will say so. Third, watch whether intra-regional trade stabilises; if Chile and Colombia continue to buy only 5–9% of their steel from their neighbours, no amount of import restriction will restore regional demand to Brazilian and Mexican mills.
The OECD's framing — that the answer lies in international coordination on subsidies rather than in individual import flows — is realistic about where the root cause sits and pessimistic about how quickly it can be addressed. The region's own numbers suggest a harder truth: the global excess capacity story explains why the steel is available, but it does not explain why Latin American buyers stopped buying from Latin American sellers. That part is domestic, and it is the part the region can actually fix.
Source & methodology +
MetalMate separates original-source evidence from summary and analysis. Verify the original source before procurement, legal or engineering decisions.
- Article status
- Published · Version 1
- Source date
- 19 November 2026
- MetalMate publication
- 19 November 2026
- Verification
- Automated evidence checks; not human reviewed

