Why the global food system is more fragile than anyone admits

On March 7, 2022 — twelve days after Russian tanks crossed the Ukrainian border — Chicago wheat futures closed at their highest level in over a decade, roughly 50 percent above their February 14 price. By late summer, the trajectory had reversed. Prices retreated. Commentary was swift: the system had absorbed the shock, substitute suppliers had mobilised, markets had self-corrected.

That price chart is consistent with two entirely different accounts of what actually happened. In the first, the food system is resilient: disruption triggers substitution, prices signal scarcity, supply adjusts. In the second, the system got lucky. A single visible disruption — the closure of Ukrainian Black Sea export terminals — was patched through six weeks of emergency diplomacy (the Black Sea Grain Initiative, assembled using Turkish leverage and Russian willingness to participate for reasons of its own) while a second, less visible disruption unfolded simultaneously: a fertilizer supply crisis driven by Russian sanctions and an unprecedented European gas price shock, which hit precisely when farmers across the world needed to buy inputs for the 2022/23 planting season. The food kept moving. Reserves ran down. Costs migrated downstream into the next growing season’s prices. The chart came back.

The price chart coming back down is evidence of neither story. It is evidence that we ran the clock out.

The architecture of an assumption

The honest case against concern goes like this: no single country produces more than roughly 18 to 19 percent of global wheat. China leads in raw production but is a negligible exporter. The system has absorbed Soviet droughts, US embargoes, Australian floods, and regional conflicts for decades. Prices spike and retreat. The record supports it, and it has earned a certain comfortable complacency.

But “distributed” and “resilient” are not the same thing. In distributed systems — power grids, financial networks, supply chains — the failure mode that cannot be managed by adding redundant nodes is correlated failure: multiple nodes failing simultaneously because they share a single underlying vulnerability. Charles Perrow identified the structural conditions for this in his 1984 book Normal Accidents: Living with High-Risk Technologies. Perrow was writing about nuclear reactors and chemical plants, but the structural mechanism — tight coupling between components, complex interfaces where failure propagates in ways invisible until they happen — applies wherever components that appear independent are exposed to common risks. The food system has these properties. What looks like diversified supply is, on examination, supply with correlated exposure. Perrow’s argument is not that complex systems inevitably fail; it’s that when they fail, the cause is rarely a single broken component. It’s the interface — the place where two systems meet under stress.

The 2008 financial crisis provides the most precise analogy available outside food systems. Mortgage-backed securities were geographically distributed across the entire United States housing market — banks held them from Maine to California, which was classified as diversification. But every security was exposed to the same underlying variable: national house prices. Nobody categorised this as concentration risk until that variable moved. When it did, geographic distribution turned out to mean nothing. The system failed everywhere at once, because the nodes looked different but the vulnerability was one.

Global grain supply is distributed across dozens of countries and hundreds of trade routes. The right question isn’t how many sources there are. It’s whether those sources share failure modes — whether the vulnerabilities that could disrupt them concentrate somewhere. What follows answers that question at three layers: the commercial infrastructure that moves grain, the geography that produces it, and the inputs that make production possible. Then it asks what the system has built to manage failure across all three at once.

Four firms, one pipeline

Start with what’s documented before reaching for what’s contested. Two of the four companies that dominate global grain trading — Cargill and Louis Dreyfus — are private firms with no public disclosure requirements for their trading positions, hedging books, or financial exposure. The other two, ADM and Bunge, are publicly listed. Together, the four are known as ABCD. Estimates of their combined share of international grain trade span a wide range: Oxfam’s 2012 “Cereal Secrets” report put the figure at up to 90 percent of global grain trade; a 2024 European Parliament study estimated the four firms handled approximately 50 to 60 percent of world trade in essential cereals, oilseeds and protein crops in 2022. That range — from roughly half to nearly all of international grain trade, depending on year, commodity scope, and methodology — is not a definitional disagreement. It is evidence that the system cannot quantify its own dependence on these four firms, because two of them aren’t required to disclose.

The opacity matters more than whatever number the market share actually is. The ABCD companies are not simply dominant traders. They own or control substantial portions of the physical infrastructure that internationally traded grain moves through before crossing an ocean: storage silos in the producing regions, grain elevators along the major river systems, port terminals at the export points. They provide the logistics networks and the hedging and financing mechanisms that make international grain trade viable for smaller participants. When these firms face simultaneous operational stress — as occurred in February and March 2022, when operations at Ukrainian Black Sea terminals curtailed — the market does not route around them. It waits.

The antitrust instinct misreads the problem. Market concentration can be disrupted by regulators. Infrastructure concentration cannot be undone without rebuilding the infrastructure. And opaque infrastructure cannot be stress-tested from the outside. The direction of consolidation is toward more concentration, not less: Bunge and Viterra completed their merger on July 2, 2025, after a regulatory process spanning two years. The ABCD-adjacent infrastructure layer contracted further.

The hidden balance sheet

During the 2022 price spike, commodity derivatives markets experienced extreme volatility. Two of the four ABCD firms were under no obligation to disclose their hedging exposure to that volatility. A major financial failure at one of these firms during simultaneous market stress would not merely disrupt its own trade flows — it could trigger cascading margin calls across the commodity finance system, extending well beyond grain. How large that exposure was in 2022 is unknowable from the outside. That is precisely the problem. The food system's most critical commercial chokepoint is also its least measurable one.

The Wheat Belt and its weather

Even if the ABCD infrastructure could somehow be bypassed — which it cannot — the supply geography it serves is itself concentrated in ways the geopolitical framing of most food security coverage tends to obscure.

Russia is the world’s single largest wheat exporter, responsible for approximately 22 to 25 percent of global wheat exports in recent years and a record 45 million metric tonnes in the 2022/23 marketing year, according to USDA Foreign Agricultural Service data. Ukraine is a major exporter. Canada’s prairie wheat belt contributes a further significant share. Together, Russia, Ukraine, and Canada represent roughly 40 to 45 percent of global wheat export volume — not a majority in the strict arithmetic sense, but a concentration substantial enough that simultaneous disruption in any two of the three would materially stress global supply.

But the percentage isn’t really the point. The point is the shared vulnerability. Russia and Ukraine are both continental Northern Hemisphere climates exposed to the same patterns of extreme summer heat. Canada’s prairie wheat belt depends on precipitation regimes increasingly subject to drought variability. And these are not independent climate risks. The major grain-producing regions of the Northern Hemisphere share exposure to the same atmospheric dynamics: jet stream disruption, Arctic amplification, the northward shift of drought-prone pressure systems. A single anomalously hot, dry Northern Hemisphere summer can simultaneously damage Russian wheat, reduce Canadian spring wheat yields, and stress US winter wheat in the same growing season. The geographic diversification that looks reassuring on a supply map does not translate into climatic independence. Different countries, but the same sky.

The 2010 Russian heat wave demonstrated what a single-country version of this looks like. The hottest summer Russia had recorded in 130 years cut wheat production by approximately a third, triggered a Russian government export ban, and sent global wheat prices to their highest level in nearly two and a half years — contributing to the food inflation that preceded the Arab Spring uprisings. That was one country, one season. The climate literature suggests the probability of simultaneous breadbasket failures across multiple Northern Hemisphere regions is already rising. Gaupp et al., writing in Nature Climate Change in 2020, estimated that the probability of simultaneous production shocks exceeding 15 percent of global grain output roughly doubles — from approximately once in a hundred years under 1998-2017 baseline conditions — to approximately once in fifty years by 2030. That is not a tail risk wheeled out for rhetorical effect. It is a documented, increasing probability with a named physical mechanism: the same atmospheric systems that govern whether Russia’s steppe regions get rain in July also govern whether Canada’s prairies do.

2010, and what came after

The pathway from Russia's 2010 heat wave to the political upheavals of 2011 is well-documented. Russian output fell sharply, an export ban followed, global wheat prices reached a 28-month high, and food import costs spiked across North Africa and the Middle East — regions where bread subsidies were already strained and household food expenditure represented a large share of disposable income. Tunisia, Egypt, Libya: the food price contribution to the conditions preceding the Arab Spring protests is established in the academic literature on food-price and conflict correlation. Food system disruption does not stay in the food system. The stakes of the structural argument in this article are not hypothetical; they have precedent.

The chemistry of hunger

If the commercial infrastructure concentrates risk at one layer, and the supply geography concentrates it at a second, the input supply system concentrates it at a third — and this is the layer fewest people have traced all the way through.

Roughly 3.8 billion people are alive today because of the Haber-Bosch process, which synthesises atmospheric nitrogen into ammonia using natural gas as the hydrogen source. Erisman et al.’s 2008 paper in Nature Geoscience on a century of ammonia synthesis established that approximately 48 percent of the global population depends on Haber-Bosch nitrogen; applied to current population, that proportion yields roughly 3.8 billion. Without synthetic nitrogen fertilizer, approximately half the current global population cannot be fed by the agriculture that remains. The process runs on natural gas. Full stop.

When gas prices spike — as they did in Europe through 2021 and into 2022, reaching record levels before and after the invasion — nitrogen fertilizer production becomes economically unviable at industrial scale. And this was not a 2022 story. It started a year earlier, from an energy price shock with nothing yet to do with the war. CF Industries halted operations at its Billingham facility in the UK in September 2021. Yara International, the world’s largest nitrogen fertilizer producer, curtailed European ammonia production by around 40 percent that same month. By the time Russia invaded, European nitrogen output was already significantly below capacity. The fertilizer crisis preceded the war. The war made it worse.

Phosphate is the second input. Morocco holds approximately 70 percent of the world’s phosphate rock reserves, according to the USGS Mineral Commodity Summaries 2024. The state mining company, OCP Group, is the world’s largest phosphate exporter. Unlike nitrogen, phosphate cannot be synthesised. It is mined from geological deposits laid down over millions of years. The accessible high-quality reserves are extraordinarily concentrated. There is no Haber-Bosch equivalent for phosphorus — no process that produces it from energy and atmosphere. A governance decision in Rabat can affect the cost of growing food on every continent. That is arithmetic, not alarmism.

Potash is the third input. Canada holds approximately 31 percent of global potash reserves, Belarus approximately 21 percent, Russia a further substantial share. Together, the three countries account for roughly 69 percent of global reserves, according to Natural Resources Canada data and the USGS. Belarus and Russia are under EU and US sanctions — Belaruskali was sanctioned in 2021 following the Lukashenko regime’s forced diversion of a Ryanair aircraft; Russian potash producers were sanctioned after February 2022. Potash prices more than tripled between early 2020 and April 2022, when muriate of potash hit an all-time high of over $1,200 per metric tonne.

All three inputs — nitrogen, phosphate, potash — hit extreme price levels in the same window. Natural gas prices drove nitrogen costs. Geopolitical sanctions drove potash. Chinese export restrictions in 2021, combined with broad commodity inflation, drove phosphate. On ordinary days, these inputs price independently, each responding to its own supply and demand dynamics. On extraordinary days — days defined by energy shocks and geopolitical disruption — the same underlying causes move all three simultaneously.

Russia's double role

Russia is simultaneously the world's largest wheat exporter and its largest nitrogen fertilizer exporter, and a significant potash producer. Sanctions applied in February 2022 cascaded across the food system at every layer at once: wheat supply disrupted, nitrogen supply disrupted, potash supply disrupted. One geopolitical trigger. Three simultaneous food system shocks. The 2022 crisis is not merely a case study in what correlated failure might look like under some future scenario. It is a documented instance of correlated failure that already occurred.

The same trigger, different faces

Everything before this section has been excavation. This is the argument.

The food system’s chokepoints — the commercial infrastructure, the supply geography, the input supply — are not three independent risk concentrations that might, in some extreme scenario, fail simultaneously. They share structural vulnerabilities that the same triggers can trip at once.

Climate extremes are the first. A single hot, dry Northern Hemisphere summer can simultaneously damage Canadian wheat, reduce US corn yields, cut Russian grain production, and stress Chinese rice-growing regions through Yangtze drought — reducing river discharge across agricultural systems that are dominant within their respective countries and regions. These are not independent events correlated only by bad timing. They are driven by shared atmospheric dynamics — jet stream disruption, Arctic amplification — that operate across regions at the same time. Gaupp et al. quantified the result: the probability of a simultaneous production shock exceeding 15 percent of global grain output roughly doubles by 2030, from approximately once in a hundred years historically to approximately once in fifty years under current climate trajectories. That increasing probability is not driven by any one country’s crop failing. It is driven by the same atmospheric systems that make the supposedly diversified supply map a single risk surface.

Energy price shocks are the second trigger. Their reach across the food system is wider than most accounts acknowledge. Natural gas prices do not only raise nitrogen fertilizer costs. They raise shipping costs, because bunker fuel prices track the broader energy environment. They raise food processing costs — industrial drying, refrigeration, packaging. They raise the cost of synthetic pesticides, which are petrochemical-derived. They raise the operational costs of the ABCD firms — storage, drying, logistics — which are passed through to prices. When energy prices spike, the entire cost architecture of the food system inflates simultaneously, not in sequence. The 2021-2022 window demonstrated this: fertilizer costs, shipping costs, and food processing costs rose together across the same twelve months.

Geopolitical disruption is the third. Russia is its clearest instance. Simultaneously the world’s largest wheat exporter and its largest nitrogen fertilizer exporter and a significant potash producer, Russia is a single point of failure across three separate layers of the food system. One sanctions regime disrupts wheat supply, nitrogen fertilizer supply, and potash supply at the same time. This is not a scenario. It happened.

The three triggers also couple with each other — which is the structural condition that makes the food system’s exposure qualitatively different from ordinary supply risk. Climate extremes drive energy demand, which drives energy prices. Energy price volatility creates geopolitical stress around energy supply, which produces supply restrictions and sanctions. Supply restrictions affect energy markets, which affect fertilizer economics. The coupling is not speculative; it was visible in the sequence from 2021 through 2022. Higher gas prices in Europe in 2021 preceded the Russian invasion and contributed to the political economy that made Russian gas supply a geopolitical instrument. The triggers do not merely co-occur. They amplify each other.

The food system has no buffer architecture designed for a scenario in which all three triggers are simultaneously active. Strategic grain reserves are sized for single-point disruptions — one country’s harvest fails, reserves cover the gap while markets adjust. The FAO’s Agricultural Market Information System, established after the 2008 food crisis to provide near-real-time production and trade data across major agricultural producers, is exactly what its name says: an information system. It shows the problem with clarity. It does not resolve it. The G7’s food security pledges are voluntary. The institutions are calibrated for the previous disaster.

Designed for the wrong disaster

That calibration is not a failure of political will. It is a structural feature of institutional risk modelling, and it deserves to be named precisely.

China holds what the USDA estimates to be the world’s largest grain reserves by volume, though Chinese authorities do not publish official figures and outside estimates vary widely. India maintains large buffer stocks. The WFP’s emergency procurement system allows rapid acquisition of food aid. The G7 Rapid Response Forum on Food Security, established in 2011 and relaunched after the 2022 crisis, provides a coordination mechanism among major producing nations.

What they cannot do is respond to simultaneous correlated failure. When all major exporting regions face simultaneous supply stress, there is no substitute supplier to mobilise — reserves run without resupply because the resupply source faces the same conditions. When all three fertilizer inputs are simultaneously unaffordable, farmers in every country face the same cost constraint at the same moment; the AMIS dashboard shows the problem in real time but cannot resolve it. When energy prices are elevated globally, the logistics infrastructure required to move emergency grain faces the same cost constraints as normal commercial trade.

The failure is not in the institutions. It is in the model the institutions are running.

The OECD-FAO Agricultural Outlook 2025-2034 — the document read by every food security policymaker with standing — describes its baseline as projecting conditions “under expected weather and yield trends, and specific macroeconomic and policy assumptions.” The Outlook does run stochastic uncertainty scenarios: 1,000 simulations in the 2025-2034 edition. But those scenarios vary macroeconomic and yield parameters independently, as separate draws from uncertainty distributions. They do not model correlated simultaneous triggering across the fertilizer input layer, the trade infrastructure layer, and the climate layer by triggers that are themselves coupled. The USDA World Agricultural Outlook Board’s annual baseline models bounded commodity-level disruptions under standard assumptions — not simultaneous multi-input, multi-exporter correlated stress. These documents shape what policymakers believe is possible. Their structural assumption is that disruptions arrive one at a time.

The Black Sea Grain Initiative is instructive not as a success story but as a ceiling — the maximum that improvisation achieved, and what it could not reach. Assembled in six weeks of emergency diplomacy in summer 2022, using Turkish leverage, UN administrative capacity, and Russian willingness to participate, it moved approximately 33 million metric tonnes in nearly twelve months, according to UN Joint Coordination Centre records. Russia terminated it on July 17, 2023. Ukrainian grain continued to move via alternative routes at higher cost and lower volume, which the institutional framework registered as a successful adaptation.

But the BGSI addressed one disruption. The simultaneous fertilizer cost crisis had no comparable improvisation available. There is no emergency diplomatic agreement that makes natural gas cheaper. There is no treaty that creates new potash mines on a usable timeline. The 2022 crisis produced one Black Sea deal. The correlated failure scenario this article describes would require five.

The model and the system

Return to the price chart. The spike and the retreat — what does it actually show?

It shows a Black Sea Grain Initiative negotiated under emergency pressure, fertilizer reserves run down rather than replaced, and farmers in import-dependent countries absorbing input cost increases that won’t surface in the global price index for another growing season. The food kept moving because every available improvisation was deployed, and the disruptions arrived in a sequence that permitted sequential response.

The food system’s actual architecture — four trading companies whose internal exposure is partially unknowable, major exporting nations concentrated in the same atmospheric hazard zones, three fertilizer inputs each controlled by a handful of countries, one energy system connecting all of them — does not guarantee that sequence. The triggers that can disrupt it are the same triggers. And they couple.

The FAO publishes food security scenarios. The USDA publishes supply and demand outlooks. The G7 convenes task forces. The institutions are, by the standards of their own models, doing their jobs.

Those models assume disruptions arrive one at a time.

The food system doesn’t.

Gen AI Disclaimer

Some contents of this page were generated and/or edited with the help of a Generative AI.

Media

MELQUIZEDEQUE ALMEIDA – Pexels

Key Sources and References

Perrow, Charles. Normal Accidents: Living with High-Risk Technologies. Basic Books, 1984.

Erisman, Jan Willem, Mark A. Sutton, James Galloway, Zbigniew Klimont, and Wilfried Winiwarter. “How a century of ammonia synthesis changed the world.” Nature Geoscience, vol. 1, no. 10, 2008, pp. 636–639. https://doi.org/10.1038/ngeo325

Gaupp, Franziska, Jim Hall, Stefan Hochrainer-Stigler, and Simon Dadson. “Changing risks of simultaneous global breadbasket failure.” Nature Climate Change, vol. 10, 2020, pp. 54–57. https://doi.org/10.1038/s41558-019-0600-z

Oxfam International. “Cereal Secrets: The world’s largest grain traders and global agriculture.” Oxfam Research Reports, August 2012. https://policy-practice.oxfam.org/resources/cereal-secrets-the-worlds-largest-grain-traders-and-global-agriculture-242474/

USDA Foreign Agricultural Service. “Grain: World Markets and Trade.” Monthly circular. https://www.fas.usda.gov/data/grain-world-markets-and-trade

USGS. Mineral Commodity Summaries 2024. U.S. Geological Survey, January 2024. https://pubs.usgs.gov/periodicals/mcs2024/mcs2024.pdf

Natural Resources Canada. “Potash facts.” Minerals and Mining Statistics, 2023. https://www.nrcan.gc.ca/our-natural-resources/minerals-mining/minerals-metals-facts/potash-facts/20521

European Parliament, Directorate-General for Internal Policies. “The role of commodity traders in shaping agricultural markets” (IPOL_STU(2024)747276_EN). 2024. https://www.europarl.europa.eu/thinktank/en/document/IPOL_STU(2024)747276

Bunge Limited. “Bunge and Viterra Complete Merger to Create Premier Global Agribusiness Solutions Company.” Press release, July 2, 2025. https://bunge.com/Press-Releases/Bunge-and-Viterra-Complete-Merger-to-Create-Premier-Global-Agribusiness-Solutions-Company

OECD and FAO. OECD-FAO Agricultural Outlook 2025-2034. OECD Publishing, Paris, 2025. https://doi.org/10.1787/601276cd-en

FAO Food Price Index. Food and Agriculture Organization of the United Nations. Monthly data. https://www.fao.org/worldfoodsituation/foodpricesindex/en/

UN Joint Coordination Centre. Black Sea Grain Initiative: Joint Coordination Centre. Data and reports, 2022–2023. https://www.un.org/en/black-sea-grain-initiative

CF Industries Holdings. “CF Industries Holdings, Inc. Commences Restart of Ammonia Plant at Billingham, UK, Complex.” Press release, September 21, 2021. https://www.cfindustries.com/newsroom/2021/billingham-restart

Yara International. “Yara Curtails Ammonia Production Due to Increased Natural Gas Prices.” Press release, September 17, 2021. https://www.yara.com/corporate-releases/yara-curtails-ammonia-production-due-to-increased-natural-gas-prices/