Researchers call it leakage: when a policy or agreement successfully reduces deforestation in the area it covers, but the underlying economic demand for cleared land doesn't shrink — it relocates to wherever the same activity remains legal, unregulated, or simply unwatched. It's a well-established finding in the peer-reviewed literature, documented across multiple journals and multiple commodities. We didn't set out looking for it. It turned up on its own, independently, in two of our last three commodity posts.
Case one: soy, and a pact now unravelling
We wrote recently about how the Amazon Soy Moratorium succeeded at what it set out to do — Amazon deforestation for soy fell sharply after 2006 — while the Cerrado, the savanna biome next door with a legal conservation requirement of just 20–35% instead of the Amazon's 80%, absorbed the displaced pressure. Multiple peer-reviewed studies describe this exact mechanism as leakage, distinct from the moratorium's real and measurable direct impact within the Amazon itself.
The moratorium itself is now under pressure. A new Mato Grosso state law strips tax breaks from companies holding suppliers to standards stricter than Brazilian law requires — the moratorium being the clearest example. Several major traders have signalled they're stepping back, with analysts warning Amazon deforestation could rise by as much as 30% as a result.
This is the sharper version of leakage: not just displacement to a neighbouring region, but a policy environment where the mechanism that prevented direct deforestation is itself being financially penalised into retreat.
Case two: palm oil's moving frontier
We found the same shape in palm oil, without needing to look for it. As Sumatra and Kalimantan — Indonesia's longest-developed oil palm regions — have come under sustained scrutiny, certification pressure and, more recently, tighter enforcement, industrial expansion has increasingly shifted toward Papua. Nearly 70% of 2025's deforestation linked to industrial oil palm expansion was concentrated there. Nobody planned this as a strategy in the way Mato Grosso's tax law is an explicit policy choice — it's the aggregate effect of thousands of individual sourcing and investment decisions responding to the same incentive: go where the scrutiny hasn't caught up yet.
Case three: wood, and the framework with a gap built in
Our wood post touched on a structural version of the same pattern. The EU's FLEGT framework governs trade with a specific list of Voluntary Partnership Agreement countries — Cameroon, Ghana, Indonesia, Liberia, the Republic of Congo, among others. Brazil, a major tropical timber supplier to the EU, sits entirely outside that framework. A governance system built country by country, through voluntary agreement, will always have gaps at its edges by design — not because any single country failed, but because the framework itself doesn't cover everywhere at once.
Why this isn't a coincidence, and why it isn't cynicism either
None of this means voluntary sustainability schemes are worthless. The Amazon Soy Moratorium prevented real, measured deforestation for nearly two decades — the academic consensus on that point is solid. What the leakage research shows is narrower and more specific: a scheme's local success is not evidence of overall success unless someone is also watching where the displaced demand went. Treating a moratorium, a certification, or a trade agreement as having "solved" deforestation for a commodity, rather than having moved the boundary of where it's still permitted, is the mistake the data keeps catching.
A boundary that works perfectly still has an outside. Leakage isn't a failure of the boundary — it's what happens to demand that was never asked to disappear, only to relocate.
Why this is the argument for EUDR's design, not against it
This is precisely the failure mode EUDR was built to avoid being. It doesn't cover one biome, one certification scheme, or one list of partner countries — it applies a single deforestation-free requirement, tied to a single cutoff date, to product entering the EU market from anywhere. In principle, there's no adjacent "not yet regulated" region for demand to move into, because the requirement travels with the product rather than staying fixed to a place. Whether enforcement in practice lives up to that design is a separate, open question — but the design itself is a direct response to what voluntary, regionally-bounded schemes have repeatedly shown they can't fully close.
Leakage is invisible until someone can actually see where a specific plot sits, region by region, plot by plot. That's what accurate geolocation makes possible in the first place — not preventing displacement, but making it visible enough to track rather than something that shows up three years later in someone else's satellite report.
A regulation that follows the product instead of the place is only as good as the location data proving where the product actually came from.
There's a second, related question worth its own piece: what happens to a shipment that fails EU due diligence when the same company also sells into the US, Asia or elsewhere with no equivalent requirement. Geographic leakage moves deforestation to the next unregulated place. Market leakage may simply move the product.
Leakage research is drawn from peer-reviewed findings in Nature Food (2020), PNAS (2012), Science Advances (2019), Land Economics (2021), Scientific Reports (2021), and PMC/Global Change Biology (2022). The February 2026 Soy Moratorium update is drawn from Mongabay's reporting on Mato Grosso's new tax law and trader exits.
← Back to Blog