We screened 13 staples across 180 countries to find whose plate breaks first when one supplier stops.
Every plate of food is the last stop on a supply chain, and some chains have exactly one link. Lesotho, a mountain kingdom of 2.3 million people, buys 100% of its imported maize from South Africa. Maize is nearly half of what a person in Lesotho eats in a day.
So we asked a simple question of the trade records the UN's Food and Agriculture Organization keeps for every country on earth: if the biggest supplier stopped tomorrow, whose plate breaks first?
of a person's daily calories in Lesotho arrive from one foreign country, through one crop.
We call that number plate share. It is easy to compute and easy to rank. Ranking it is where the argument started.
| Country | Crop · via | Plate share | Pop (M) | Full plates (M) |
|---|
The fifteen countries with the highest plate share. Each bar is the share of one person's daily calories that comes from one foreign supplier through one crop. Lesotho leads. Nothing is highlighted yet.
Watch which bars stay blue. Every one is a country of fewer than 12 million people. Wheat dominates, and the suppliers are Russia, Kazakhstan, Ukraine, Serbia. Lebanon's 21% via Ukraine was visible here before February 2022.
Now change the question from how deep to how many: multiply plate share by population. The bars re-sort. Pakistan, Mexico and Egypt rise; Lesotho falls to the bottom. Same data, opposite answer.
The orange bars are the three we followed. 14 million people's worth of full plates in Mexico hang on one supplier, the United States. That is the scale a national budget answers to.
The obvious next step was to simulate the shock: take the top supplier away, let the others fill the gap up to the most they had ever shipped, and count the tonnes that could not be replaced. Convert tonnes to calories. Divide by people.
The answer, for every country, was zero. Brazil had spare maize for Mexico. Australia had spare wheat for Egypt. On paper, the ships always come.
Before trusting that, we did the one thing a model can't argue with: we checked it against something that had already happened.
| Year | Volume (2019 = 100) | Price per tonne (2019 = 100) |
|---|
Egypt imports more wheat than any country on earth. Here is how much arrived each year, indexed so 2019 is 100. The grey line marks February 2022, when Russia invaded Ukraine. Together they supplied three-quarters of it.
Now the price of each tonne, on the same scale. It climbs through 2022 and stays high for two years. The ships kept coming. What arrived on the bill did not.
Average the three years after against the five before. Volume: −3%. Price: +54%.
Egypt paid. It bought 3% less wheat and paid 54% more for it. But when we ran the same before-and-after for our other cases, they didn't follow Egypt's script.
Mexico paid more and bought more. Its maize imports rose 18% through a 36% price rise, and its supplier count fell from 24 to 8. The shock made it lean in harder.
Ghana couldn't pay. Its rice price barely moved in dollars, but the cedi lost half its value in 2022 and the country defaulted that December. Ghana bought 29% less rice. Not because rice got dearer; because Ghana got poorer.
| Country · crop | Volume | Unit price | Mechanism |
|---|---|---|---|
| Egypt · wheat | −3.1% | +54.0% | paid more |
| Mexico · maize | +18.5% | +35.8% | paid more, bought more |
| Ghana · rice | −29.4% | −3.8% | couldn't pay, bought less |
Same fragility, three different endings. What decides the ending is not the crop, the supplier, or the distance. It is whether the country can afford the bill. That is the thing our plate-share number never measured.
We rebuilt the model around price. If the market reprices the way it did for Egypt in 2022, and a country's food budget does not grow to match, this is what goes missing from each person's plate every day:
per person per day in Egypt, about two slices of bread, for 111 million people. Mexico: 137. Ghana: 78.
A planner has two instruments. Hold grain, which costs money every single year in silos, interest and spoilage. Or hold a budget line, which costs nothing until the shock arrives and then pays the higher bill. Our first draft sized a reserve with one formula and got the same answer for all three countries, 4.2 months. That is the model read back, not a decision.
So we costed the two instruments against each other. The ratio between a shock-year bill and a year of carrying grain is the break-even interval: a reserve only pays for itself if shocks come more often than that. Give the tool a country and a price shock and it returns both numbers, and the interval.
Choose a case and a price shock. Egypt's observed 2022 shock is the default.
Bounded: 0 kcal if fully replaced at the same price · 291 kcal if nothing is replaced. Model C: imported calories × (1 − 1 ÷ (1 + shock)).
Carry assumes USD 20 per tonne a year for storage, 5% capital cost and 2% shrinkage on grain valued at the 2019–23 import price. These are typical bulk-grain figures, not FAOSTAT data; halving or doubling storage moves the break-even by a few years, not the conclusion.
Which instrument depends on how the plate broke. A country that paid more needs money. A country that leaned in on a corridor with no alternative needs its own stock. A country that could not pay needs hard currency, not grain. One rule per country, each with a trigger, a size, a yearly cost and an owner.
Trigger: imported wheat above about USD 350 a tonne, the 2017–21 average plus half the 2022 shock, for two quarters running. Action: release a USD 1.7 billion budget line, not the silos. We checked this against the event. Egypt paid USD 1.59 billion, 1.52 billion and 0.82 billion above 2017–21 prices in 2022, 2023 and 2024, so the line covers both peak years with about 5% to spare and over-covers the average year by a quarter, which is the right shape for a contingency. Prices stayed above the trigger for two years, and a 4.2-month reserve would have bridged one of them. Carrying that reserve costs USD 0.14 billion a year and only pays off if a shock like 2022 recurs more than once every 12 years. Sign forward contracts now with the suppliers that stepped up: France, Bulgaria, Romania. Owner: GASC with the Ministry of Supply.
Trigger: the US share above 90% for two years (it is 94% now), or the maize import price above about USD 270 a tonne, a quarter over its 2017–21 average. Action: a domestic strategic stock of 6 million tonnes, 4.2 months of imports. Here money cannot buy supply: no other supplier has ever shown more than 0.6 points of extra capacity, and the rail and ports from Brazil and Argentina do not exist yet. Cost: about USD 1.6 billion to fill once, USD 0.23 billion a year to carry, against a USD 2.4 billion bill in a shock year. The carry is the insurance premium for a single-corridor supply. Owner: SEGALMEX with SADER.
Trigger: quarterly rice imports below 85% of the trailing three-year average, or the cedi down more than 20% in a year. Both happened in 2022–23. Action: a hard-currency import line of USD 0.11 billion a year, the 29% of tonnes Ghana stopped buying valued at base prices, released on the trigger. Not a grain reserve: the price never moved, the constraint was dollars. Rebuild Thailand, 16% of the market before 2023 and 3% after, as the deliberate second supplier. Owner: Bank of Ghana with the Ministry of Food and Agriculture.
Across all three, a year of carrying grain costs about a tenth of a shock-year bill, so a budget line is the cheaper first instrument unless shocks arrive more than once a decade. Grain reserves earn their keep where money cannot buy supply.
We went looking for the country that runs out of food. We found that almost nobody does. The ships come. Then the bill comes. And whether a plate breaks depends on who is holding it.
Sources: FAOSTAT Detailed Trade Matrix · FAOSTAT Food Balances (2010–). Analysis notebook and code available from the team.