Colombia Soybean Grain Market Size, Share & Forecast 2026–2033

What this market is

Colombia's soybean grain market is the value of the whole soybeans the country buys from abroad and converts — through crushing, refining and feed milling — into the protein behind its poultry, pork, dairy and aquaculture sectors, together with the cooking oil in its kitchens. Colombia grows only a small fraction of the soybean it consumes; the country's protein economy runs on imported beans and imported meal.

Pheonix puts that import pool at 1,420 thousand tonnes in 2025, worth USD 632 million at an average landed unit value of USD 445 per tonne. On the same model the pool reaches 2,760 thousand tonnes and USD 1.31 billion by 2033 — a volume CAGR of 8.7 per cent and a value CAGR of 9.5 per cent. Price contributes only a small part of that: revenue per tonne rises from USD 445 to USD 475 across the window, or 6.6 per cent over eight years, roughly 0.8 per cent a year. Almost all of the value created is tonnage, not price.

The number that anchors this report

In 2025 Colombia imported 34 per cent more soybean grain than in 2024. Worked backwards through that growth rate, the 2024 pool was approximately 1,060 thousand tonnes. The obvious reading is that 2025 was a restocking spike and that the line mean-reverts — and that reading is the one this report argues against, for a reason that shows up in the arithmetic rather than in the trade statistics.

A spike year produces a jump in volume and a fall in unit value, because the marginal tonne is bought when supply is long and prices are weak. This one did something else. Colombia's soybean imports rose 34 per cent in a year in which the global bean complex was oversupplied by a record South American crop, and the value of the pool rose with the volume rather than against it. A buyer filling a warehouse behaves that way. A buyer switching the composition of what it buys — substituting imported beans for imported meal, and crushing them at home — also behaves that way, and keeps behaving that way.

Why 34 per cent is a share shift, not a spike

Colombia's animal-protein economy is the largest in the northern Andes and the least self-sufficient in feed grain. The country's compound feed industry sits inside the 47.759 million tonnes Alltech's 2026 Agri-Food Outlook attributes to the residual of Latin American markets outside Brazil, Mexico, Argentina and Ecuador — a bucket that runs from Colombia and Peru through Central America and the Caribbean. That regional tonnage is growing, but not at 34 per cent; the whole of Latin America fed only 5.5 million tonnes more in 2025 than in 2024, on a base of 204.4 million. Colombian soybean imports cannot be explained by Colombian feed demand alone. Something else moved the composition.

What moved is the crush. Whole beans carry the oil leg; soybean meal does not. When the oil leg is valuable, a processor who imports beans captures it in Colombia rather than leaving it with the exporter. And the oil leg has been the strong half of the complex globally: throughout 2024–2025, soybean oil's value as a biofuel feedstock pulled crush margins in a way that meal demand never could, which is why US crush ran to a record 2.49 billion bushels for 2025/26 with oil, not meal, setting the marginal economics. A Colombian refiner facing a domestic biodiesel blend mandate and a cooking-oil market with real brand premium sees the same arithmetic an Iowa processor sees. Importing the bean is how you take the oil.

That gives the 34 per cent a mechanism rather than a mood, and it also explains why the growth should not fully revert. Bean share rises while crush capacity and refining capacity are being added; it plateaus when they are built out. The forecast below assumes the plateau is reached gradually across the window rather than in 2026.

The comparison that keeps the size in proportion: the same USDA Foreign Agricultural Service post reporting that carries Colombia's feed outlook puts Colombian corn imports at 7.2 million tonnes in MY 2025/26, against domestic corn production of 1.55 million tonnes. Soybean grain at 1,420 thousand tonnes is 19.7 per cent of the corn import tonnage. Colombia is a feed-deficit country that buys its energy grain and its protein separately, on different contracts, from different origins, at different points on the price curve — and that separation is exactly why the protein side can re-base without the energy side moving.

Market segmentation

Four cuts matter for anyone selling into this market: who consumes the protein, where the bean comes from, which port it lands at, and who signs the purchase order. All shares below are Pheonix estimates for 2025, applied to the 1,420 thousand tonne pool.

By end-use sector

By product stream — the technical split of a crushed tonne

By origin

By port of entry

By buyer type

Basis of estimate — read this before quoting a number

The growth path in this report is anchored on one hard datum: the 34 per cent increase in Colombia's soybean grain imports in 2025. The absolute base — 1,420 thousand tonnes at USD 445 per tonne — is a Pheonix bottom-up estimate built from feed-sector protein requirements, crush yields and landed unit values, not a line lifted from a customs series. The same is true of every percentage share in the segmentation tables above. They are modelled positions on a market whose published trade detail is thinner than its importance warrants, and they should be read as directional and internally consistent rather than as declarations.

Two external anchors are used and are identified in the text where they appear: USDA Foreign Agricultural Service post reporting on Colombia's grain balance (corn imports 7.2 million tonnes, production 1.55 million tonnes, consumption 8.68 million tonnes for MY 2025/26, with Central Bank growth of 2.6 per cent in 2025 and 3.4 per cent in 2026), and Alltech's 2026 Agri-Food Outlook for regional feed tonnage (204.4 million tonnes across Latin America, up 5.5 million). Where a reader needs a customs-certified base rather than a modelled one, the segmentation is the part to replace first: the growth story survives a re-basing, the shares do not.

Connected Pheonix research

• Latin America Animal Feed Market Size, Share & Forecast 2026–2033 — the regional feed demand this import pool feeds into
• Global Animal Feed Market 2026–2033 — mill economics and least-cost formulation at global scale
• Global Sunflower Seeds Market 2026–2033 — the other oilseed Colombia's refiners can substitute into
• Global Feed Phosphates Market 2026–2033 — the mineral half of the same ration
• Global Seed Market 2026–2033 — the input side of the South American crop that supplies Colombia
• Global Insect Protein Market 2026–2033 and Black Soldier Fly Market 2026–2033 — the substitution threat at the margin of the protein fraction
• Global Glyphosate Market 2026–2033 — the crop-protection input that determines South American bean cost of production
• Global Seaweed Extracts Biostimulant Market 2026–2033 — the crop-input side of the South American yield story that sets Colombian bean cost

Table of Contents

• Value chain & crush economics
• Technology & logistics
• Competitive landscape
• Regulatory & policy
• Investment & capacity
• Market risks & scenarios

Competitive Landscape

Competitive landscape

The market is moderately consolidated at the top and fragmented underneath. Five actors matter, and only two of them are Colombian-headquartered.

The structural point is that the two global traders do not compete with the two Colombian processors for the same margin. Bunge and Cargill earn on origination, freight and risk transformation — they are the counterparty on the other side of the FOB contract, and their Colombian presence is a service to that book. Team Foods and Solla earn on formulation, brand and distribution. The line between the two models is moving, though, in one direction: when an integrated protein producer decides to import beans and toll-crush rather than buy finished meal, it takes the trader’s spread and the crusher’s spread in the same decision. That is the competitive event to watch in this market, and it does not show up in market-share tables.

Consolidation has already reshaped the supplier side. Bunge’s completed combination with Viterra created a materially larger origination footprint across exactly the origins that supply Colombia — Brazil, Argentina, Paraguay — which narrows the number of counterparties able to offer a full-programme contract across multiple origins and grades. For a Colombian buyer, fewer full-programme sellers means less optionality on payment terms and freight allocation, even if the underlying commodity stays perfectly competitive on price.

Function by function, who captures what

The reason market-share tables mislead in this market is that the participants are not competing for the same pool of margin. They are paid for different functions, and a change in structure reallocates margin between functions without moving a single tonne.

Vertical integration in this market is best understood as a decision about which of those six lines to own. A trader that buys a crush plant is buying the fourth line and keeping the first two. An integrated protein producer that starts importing beans and toll-crushing is buying the fourth line while trying to surrender the first two — a structurally different and harder trade, because origination is a book business that rewards scale and the producer’s book is one country deep.

The buyer side, in tonnes

Demand for the pool is not a single curve. On the model’s 1,420 thousand tonne base, the four buyer classes behave differently enough that they are worth sizing separately:

The 539.6 thousand tonnes sitting with integrated producers is the number that decides the competitive outcome. It is the only bloc large enough to justify a dedicated bean-supply programme, and it is the bloc whose sourcing decision converts a trader’s margin into a crusher’s margin. Read the market as two thirds price-takers and one strategic buyer and the consolidation question answers itself.

Two entry routes, and why neither is cheap

There is no low-capital way into this market. The first route is asset entry — build or buy crush and refining capacity — which requires clearing the utilisation arithmetic set out in the investment chapter, on a pool where a single modern plant represents 46.5 per cent of today’s national tonnage. The second is book entry: become a full-programme originator across the US, Brazil, Argentina and Paraguay simultaneously, absorb the freight and currency risk, and earn the basis. That requires balance sheet and a multi-origin desk, which is why the field at the top is short. A third route — buying a Colombian feed or oils platform for its customer relationships — buys the sixth line of the table, not the first four, and leaves the acquirer still buying its exposure from someone else.

Further reading: Global Animal Feed 2026–2033 and The Meat Merger That Was Really a Feed Order.

Value Chain

Value chain and crush economics

The chain that turns a Brazilian or US bean into a Colombian chicken has six links, and the margin sits in a different place at each one.

Two features of that chain are specific to Colombia. First, the country’s refining capacity is concentrated near the Caribbean coast and in the Cauca valley, while a large share of the poultry flock sits in Santander and Cundinamarca — so the internal freight leg between port, plant and farm is longer, relative to cargo value, than in Argentina or Brazil. Second, the crush spread is not the only thing the processor is managing; a domestic biodiesel blending mandate gives the oil leg an administered floor that meal can never match, which is precisely what makes importing beans more attractive than importing meal.

The ratio that matters most is the milk-to-bean relationship in the feed ration. Every point by which the price of soybean meal rises relative to corn pushes Colombian mills toward least-cost reformulation, substituting wheat, rice bran or distillers’ grains and shaving the protein fraction. At 1,420 thousand tonnes, a one percentage point change in the national protein inclusion rate is worth roughly 14 thousand tonnes of beans. That sensitivity is why the segment’s growth is not a straight line, and why the Latin America Animal Feed report is the best read-across for anyone modelling this market. See also Latin America Animal Feed 2026–2033 and Global Sunflower Seeds 2026–2033 for the substitution alternatives at either end of the crush.

Reading the pool as two multipliers, not one number

The headline 9.54 per cent value CAGR is almost entirely a tonnage number wearing a value label. Multiply the volume path by the price path — 1.0866 by 1.008 — and the product is 1.0953. Of the nine and a half points of annual growth in this pool, 8.66 come from kilos, 0.80 from price, and the interaction accounts for the remainder to two decimal places. The consequence is practical: any argument about the size of the 2033 market is really an argument about how much tonnage clears the ports, and any argument that leans on price is leaning on a component worth a twelfth of the total.

The product stream, and why the oil leg is disproportionate

A tonne of beans does not leave the plant as a tonne of saleable product. The stream is roughly 78 per cent meal, 18 per cent crude oil and 4 per cent hulls and process loss, and those fractions do not carry the same value per tonne. Oil’s share of crush revenue is therefore determined by the oil-to-meal price ratio, which is the one variable a Colombian processor cannot set:

Read down that column and the whole share-shift thesis becomes conditional rather than permanent. Importing beans instead of meal buys exposure to a fraction that is worth 19 per cent of the plant’s output at one ratio and 41 per cent at another. When the ratio compresses, a Colombian crusher that has committed to beans is worse off than one that stayed on imported meal — and no feed-demand forecast will tell you that has happened. The ratio moves before the tonnage does.

Where the chain leaks value before the crush starts

Two losses sit upstream of the plant and neither appears in a crush margin. The first is shrink — the gap between what is discharged at the port and what is weighed at the mill. In a humid climate without aerated storage, a percentage point or two is ordinary. One point on 1,420 thousand tonnes is 14.2 thousand tonnes, which at the model’s USD 445 a tonne is USD 6.3 million a year: about a fifth of the entire value increment the pool adds in a single year of its forecast path. The second is the internal haul. Refining is concentrated on the Caribbean coast and in the Cauca valley while the flock sits in Santander and Cundinamarca, so the bean and the meal each cover a domestic leg that a Brazilian or Argentinian chain would not pay for at the same relative cost.

Both leaks are fixed-cost problems with cheap answers — silo, aeration, a discharge rate, a better sited mill — which is why the investment chapter argues that storage should be sequenced before crush. The value chain does not lose money at the sophisticated end of the process. It loses it in the ordinary movement of a commodity that weighs twice what it is worth.

Further reading: Feed Is 62.5 Per Cent of the Bill. That Is Why the Mill Cannot Raise Its Price.

Investment Activity

Investment and capacity

Where the money is going in this market is crush, storage and port throughput — in that order of certainty, and in the reverse order of ease.

The arithmetic that governs all of it is simple. The pool adds USD 678 million of value between 2025 and 2033, and almost all of that increment is tonnage that has to be physically handled. A crush plant is worth building when its share of that tonnage clears the capital charge on 1,000 to 3,000 tonnes a day of extraction capacity plus downstream refining; it is not worth building on the 4 per cent scenario in which the pool never re-bases away from imported meal. That is why the investment case and the share-shift thesis are the same argument seen from two sides.

On the corporate side, two transactions define the current cycle. Bunge’s combination with Viterra concentrated origination capacity across the South American origins that supply Colombia. And on the domestic side the Colombian oils and fats industry has been consolidating into larger refining groups with consumer brands attached — a consolidation that is fundamentally about owning the retail oil premium rather than about bean throughput. Neither transaction changes the tonnage. Both change who captures the margin on it.

One modern plant is nearly half of today’s entire market

Scale arithmetic settles most of the investment argument before strategy gets a word. A plant running 2,000 tonnes a day for 330 operating days handles 660 thousand tonnes a year. Against the 2025 pool of 1,420 thousand tonnes that is 46.5 per cent of national bean flow through one asset. Against the 2033 pool it is 23.9 per cent. There is no such thing as a marginal entry into this market: the smallest economically rational crush asset is a structural event, which is why capacity arrives in lumps and why utilisation — not technology — is the risk that decides whether the capital earns.

How many plants the 2033 pool actually supports

Run the same arithmetic the other way. At the modern scale band of 1,000 to 3,000 tonnes a day — 330 to 990 thousand tonnes a year — the 2033 pool of 2,760 thousand tonnes supports:

Between three and eight plants, and the band is narrow in both directions. Fewer than three at the top of the scale means the country is exporting its crush margin; more than eight means the smallest units are running at sub-scale and will be consolidated or idled. That is a genuinely useful constraint to hold in mind when a capacity announcement is made: it either fits inside the eight or it displaces one that already exists.

Value per added tonne — the number to test capex against

The pool adds USD 678 million of value between 2025 and 2033, against 1,340 thousand tonnes of added volume. That is USD 506 of added value per added tonne of throughput. Any capacity decision can be tested against it directly: if the all-in cost of building and operating a tonne of annual handling and extraction capacity, discounted over its life, sits below USD 506, the asset clears; above it, the project is underwriting a share shift that has not happened yet. The number is an average across a chain with very different capital intensity at each link, so it belongs to the combined crush-and-handling investment, not to any single piece of equipment.

Sequencing: storage, then discharge, then crush, then refining

The capital intensity of each link and the payback logic point in the same order, and it is not the order in which announcements are usually made:

The sequence matters because the early links are cheap and protect value that the later links assume. Building extraction capacity on top of unaerated storage is a way of buying beans and then losing a percentage point of them before they are crushed — the capital is committed to the sophisticated end while the ordinary end leaks.

The utilisation trap

In the feed-growth-only case the 2033 pool reaches USD 864.9 million, which at USD 475 a tonne is roughly 1,821 thousand tonnes. Commission one 660 thousand tonne plant against the base case and deliver it into that case, and the single asset is absorbing 36.2 per cent of the entire national pool to run at its own design rate. That is the arithmetic behind the report’s central claim that this market’s upside is a capacity decision: build the plant and the tonnage must follow it, because if it does not, the plant does not idle gracefully — it re-prices the crush spread for everyone else in the country.

Further reading: Latin America Animal Feed 2026–2033.

Technology & Innovation

Technology and logistics

Nothing in this market is technologically novel, and that is itself the finding. Solvent extraction with hexane is the incumbent process worldwide; mechanical pressing survives only in small regional plants where the cake commands a food premium. Colombia’s technology question is not how to crush but where, and how fast cargo can be moved inland once it lands.

The four decisions that actually move cost

• Extraction scale. A modern plant needs roughly 1,000 to 3,000 tonnes a day to earn its capital. Colombia’s bean pool of 1,420 thousand tonnes is enough to support a handful of plants at that scale and not many more — which is why the crush build-out question in this report is a question about utilisation, not about technology.
• Port discharge rate. Pneumatic unloaders versus grab cranes on a handymax is the difference between a two-day and a five-day discharge, and demurrage on a five-day overrun is a direct subtraction from the crush margin.
• Silo and aeration. Colombian humidity makes stored beans a quality risk within weeks. Storage that is not aerated converts a cheap purchase into a distressed sale.
• Least-cost formulation software. On the feed side, formulation is the technology. A mill running modern linear-programming software re-optimises the ration every time the meal-to-corn ratio moves; a mill running a fixed formula does not.

Downstream, extrusion is where the interesting product innovation sits — textured soy protein, high-moisture analogues, full-fat soy for aquafeed — but it is a small fraction of Colombian volume and should not be modelled as a growth driver for the bean pool. The strongest read on that boundary is Textured Soy Protein Is a Crush Story, Not a Burger Story, and the substitution threat at the far end of the protein fraction is covered in Global Insect Protein 2026–2033.

What a day of anchorage costs, per tonne

Port economics are usually discussed in the abstract. They are not abstract at the level of a single cargo. On a 40,000 tonne handymax parcel, each day of delay costs the following per tonne of cargo:

The per-tonne numbers look trivial against a USD 445 cargo — half a dollar is eleven basis points. That is the wrong comparison. The number that matters is the one in the third column, because a port constraint is not a one-cargo event; it is a standing condition that every cargo pays. USD 0.50 a tonne of avoidable port cost is USD 710,000 a year, and it lands on the trader before it lands on the mill, which means it is priced into the basis the mill is quoted. Port inefficiency is not absorbed. It is passed.

What aeration is worth, and why it beats sophistication

Compare the two upgrade classes honestly. A percentage point of storage shrink on the national pool is 14.2 thousand tonnes, USD 6.3 million at USD 445 a tonne. A percentage point of avoidable port cost, in the table above, is under a tenth of that. Handling and quality preservation is where the money is, and the technology required is forty-year-old engineering: silo, aeration, moisture monitoring, fumigation. There is no innovation premium to be earned here, which is exactly why it is chronically under-invested relative to extraction capacity.

Technology, by maturity and by whether it moves the forecast

The pattern in that table is the useful part. Every technology that moves the pool is either old, cheap and physical, or old, cheap and analytical. The newer processes — extrusion, high-moisture analogues, precision fermentation routes to protein — are genuinely interesting and are covered elsewhere in this library, but they act on a few per cent of Colombian volume and should not be modelled as a driver of bean demand. A forecast that leans on the novel end of this table will be wrong about the tonnage while being right about the direction.

Further reading: Global Seed Market 2026–2033 — the upstream technology that sets South American bean cost of production.

Market Risk

Market risks and the three scenarios

The forecast has one dominant sensitivity and several secondary ones, and it is worth separating them because they are routinely blurred into a single risk sentence.

The scenario band

The gap between the outer cases is USD 757 million on the same feed demand. That is the single most useful fact in this report: the market’s upside is a capacity decision, not a consumption forecast.

Four risks that can break the base case

• Origin concentration. The top two origins supply 63 per cent of the pool. A phytosanitary suspension, a freight disruption through Panama, or a tariff change on any one of them moves delivered cost immediately and cannot be offset inside a shipping cycle.
• Currency. Beans are priced in dollars and sold into pesos. A depreciation of the peso raises the local cost of the ration faster than it raises the price of chicken, and Colombian protein producers cannot pass it through at retail without losing volume. The 2025 layer segment already showed what margin pressure does to feed demand.
• Substitution in the ration. Corn at 7.2 million tonnes of imports against 1.42 million tonnes of beans sets the ratio the formulator watches. Wheat, rice bran and distillers’ grains can each take share from soybean meal at the margin; palm oil can take share from soybean oil in refining. Neither substitution is large in a normal year, and both are fast in an abnormal one.
• Animal disease. Salmonella pressure hit Colombia’s layer sector in 2025, alongside an avian influenza cycle that reshaped layer economics across the region. A disease event does not reduce the number of birds smoothly; it reduces them in a step, and feed tonnage follows within a quarter.

The three falsifiers

Three observations would tell you the thesis is wrong before the market size does. First, Colombian bean imports falling back toward 1,060 thousand tonnes while meal imports hold — that would prove 2025 was a restock. Second, crush utilisation in the Cauca valley and around Cartagena failing to rise as tonnage rises — that would mean the beans are being re-exported or stored rather than converted. Third, the delivered premium on compliant South American beans widening beyond the value of the oil leg — that would reverse the substitution without any change in Colombian demand. None of the three is visible in an annual tonnage figure; all three are visible quarterly.

What one per cent is worth

Risk sections tend to list uncertainties without pricing them. Priced, the hierarchy is stark:

Two conclusions follow. First, the growth path dominates the starting level in the middle of the window but the starting level dominates it at the end — which is not a paradox, it is what compounding does to a scaling error. Second, and more useful: price risk is genuinely small here. A reader who spends their attention on bean price volatility is watching the least consequential variable in the table.

The slow case, in physical terms

The feed-growth-only scenario looks like a softer version of the base case until it is converted into tonnes. USD 864.9 million at USD 475 a tonne is 1,821 thousand tonnes — an increment of roughly 401 thousand tonnes over eight years, about 50 thousand tonnes a year. One modern 2,000 tonne-a-day plant consumes 660 thousand tonnes a year. In the slow case, the entire national increment over a full year is a thirteenth of one plant’s annual throughput.

That is the argument for treating the two outer scenarios as different markets rather than different forecasts. At 4 per cent, no new extraction capacity is justified on demand growth alone and the crush build-out that the share-shift thesis predicts simply does not get financed. The scenario is not a mild disappointment. It is the negation of the investment case.

Leading indicators, and how far ahead they speak

The three falsifiers in the previous section say what would disprove the thesis. This table says when you would know, and it is ordered by lead time rather than by importance:

Nothing on that list is an annual statistic. Every one of them is observable inside a quarter, which is the practical case for not treating a customs series as the instrument you manage this market with. By the time the annual import figure confirms or refutes the thesis, the crush spread has already moved and the capital has already been committed.

What would change this forecast

Two developments would shift the base case rather than merely test it. A sustained compression of the oil-to-meal ratio below the level at which domestic crushing earns its capital would convert the share shift back into a restock — the model’s 4 per cent path, arrived at through the product stream rather than through feed demand. And a phytosanitary suspension affecting either of the two origins that supply 63 per cent of the pool — roughly 894.6 thousand tonnes carrying about USD 398 million of landed value — would compress volume for a season through a channel that no demand-side forecast can anticipate. Neither is in the base case. Both are priceable, and both are visible before the annual tonnage report.

Further reading: Brazil Is 44 Per Cent of the Region and Mexico Buys Its Corn in Dollars and Black Soldier Fly Market 2026–2033.

Regulatory Landscape

Regulatory and policy landscape

Regulation in this market does not set demand; it sets the delivered cost of a tonne and the speed at which it can land. Four regimes do the work.

Phytosanitary admission

Every soybean consignment enters under an import authorisation from ICA, Colombia’s agricultural institute, tied to the pest-risk status of the origin and the processing establishment. This is the binding operational constraint on adding a new origin: the paperwork is not difficult, but the risk assessment is slow, and it is the reason a cheaper new supplier cannot be switched on inside a shipping cycle. Approval pathways in this market are compliance-driven rather than licence-based — the product is standard, the admission is not.

Tariff and trade agreement structure

Colombia’s tariff treatment of soybeans is the single largest determinant of origin share. Preferential access under the trade agreement with the United States is why US beans hold the largest single share of the pool despite Brazil’s price leadership on a record crop. Andean Community arrangements and bilateral terms with Mercosur partners govern the rest. Colombia also operates an Andean price-band mechanism across a defined list of agricultural products; where a commodity sits inside that list, the effective duty moves with international reference prices, which is a form of protection that bites hardest exactly when world prices fall — the moment importers would otherwise buy most.

Food safety and the oil leg

Soybean oil destined for human consumption is registered and inspected under INVIMA, Colombia’s food and drug regulator. This matters to the crush economics rather than to the bean trade: a refiner selling branded cooking oil carries a compliance cost that a refiner selling only into biodiesel does not, and that difference is why the same tonne of crude oil can be worth materially different amounts to two buyers at the same port.

Biofuel policy and origin traceability

Colombia’s biodiesel blending mandate is the administered floor under the oil leg, and therefore the mechanism that makes domestic crushing worth doing at all. Any change to the blend level resets the value of a crushed tonne without touching the feed side. Layered onto that is the external regime: deforestation-free sourcing requirements in export markets are pushing South American origins toward monitored, satellite-verified supply chains. Colombia is a buyer rather than a seller in this market, so those rules reach it as an input cost and an origin-selection criterion — a compliant Brazilian tonne costs more than a non-compliant one, and the premium is paid at the port of discharge.

What a percentage point of protection is worth

Tariff treatment is usually described qualitatively, which makes it hard to weigh against the freight and basis moves it competes with. Priced against the model’s USD 445 a tonne, a tariff-equivalent of one percentage point is USD 4.45 a tonne, which across 1,420 thousand tonnes is USD 6.3 million:

Read it as a tariff-equivalent rather than a duty schedule. Colombia’s instrument across much of this product list is the Andean price band, which moves the effective rate with international reference prices — so the numbers above are what a given level of protection would be worth if applied flat. The practical implication is the same and worth stating plainly: protection of five points or more is large relative to the year-to-year moves in freight and basis that importers spend their time managing. Origin share in this market is a policy variable first and a commercial one second.

The compliance year, as a sequence

The regimes described above do not act simultaneously; they act in an order, and the order determines how fast a sourcing decision can be executed:

Two features of that sequence matter more than any individual row. The first is that the two slowest steps — origin assessment and traceability infrastructure — are the two that cannot be bought with money at short notice, which is why origin concentration in this market is sticky rather than opportunistic. The second is that the blend level sits outside the sequence entirely: it is revised administratively and takes effect immediately, and it resets the value of the oil leg without any of the other four steps moving.

The traceability premium has a ceiling

Deforestation-free sourcing requirements reach Colombia as an input cost rather than an export obligation, because Colombia buys beans rather than sells them. That creates a temptation to treat the premium as an open-ended cost that can only rise — and it cannot. A premium can only be charged up to the value of the thing it protects. The switch from imported meal to imported beans was justified by the oil leg; therefore the premium on a compliant tonne can rise until it consumes the oil-leg advantage, and no further. Beyond that point the ration reverts to imported meal and the share shift unwinds without any change in Colombian demand, without a tariff, and without a phytosanitary event.

That is worth holding alongside the third falsifier in the risk chapter, because it is the same mechanism seen from the regulatory side. A compliance regime is not only a cost line. It is a live constraint on the ceiling of the substitution thesis, and the number to watch is not the premium itself but the premium expressed as a share of the oil-leg value differential.

Two variables in the rules that nobody models

The first is the cut-off date embedded in any sourcing rule, which determines how much of an origin’s existing production area qualifies. Two regimes with identical stringency and different dates have completely different effects on which region of Brazil or Paraguay can supply a compliant tonne. The second is the set of approved biotechnology events for import and use: a change to it re-prices entire origins at a stroke, because a cargo that cannot legally be used has no value to a crusher regardless of its price. Neither variable appears in a demand forecast, and either can move origin share by more than a tariff step does.

Further reading: Global Glyphosate Market 2026–2033 — the crop-protection input whose regulation shapes South American cost of production, and Global Feed Phosphates 2026–2033 for the mineral side of the same regulatory perimeter.