How Rails Fall Apart in Crisis: Understanding the Structural Failures of the Cross-Border FX Market

Harrison Mann,
Head of Growth

Introduction
Any vessel who enters will be set on fire.
This was the threat issued by Iran's Revolutionary Guard on March 2, 2026, four days after US and Israeli strikes killed Supreme Leader Khamenei. It was not an idle one, drone strikes hit tankers near the strait -- causing insurers to immediately withdraw coverage. Within 72 hours, traffic through the Strait of Hormuz -- 21 miles of water carrying 20% of the world's oil, dropped to zero.
By March 9th, Brent crude touched $119 a barrel, a price not seen since the invasion of Ukraine. Iraq and Kuwait, with nowhere to put the oil they couldn't ship, began shutting down wells. Analysts compared it to the 1973 embargo, "This is about as wrong as things could go at any single point of failure," one energy researcher told NPR.
But oil is only the commodity that makes headlines. The Strait of Hormuz sits at one end of a region that runs on a different export: labor.
Nine million Indians. Two and a half million Filipinos. Bangladeshis, Pakistanis, Nepalis, Indonesians -- 24 million migrant workers who build the infrastructure and staff the hospitals and ultimately wire their wages home. When the airspace closed on February 28th and 250,000 travelers found themselves stranded in terminals, many of these workers were stranded too, in a different way. The remittance corridors that sustain tens of millions of families who have never seen the Gulf but depend on it utterly -- those corridors went dark.
The FX market doesn't fail the way a machine does, all at once and often for a single reason. It fails the way a body fails: one organ first, then another, then a cascade that reveals how deeply the systems depend on each other.
Money stops and the people closest to the margin discover they were standing on infrastructure that was never as solid as it looked.
The Strait of Hormuz in March 2026 is the most complete expression of vulnerabilities that have been surfacing in cross-border finance, failure after failure, for over a decade. To understand why the system breaks this way, requires starting somewhere else.
Thirteen years earlier. A small island in the Mediterranean. A bank that had just confiscated half of its depositors' money.
Cyprus, 2013: The invention of a two-tier euro
Angelos was seventy-seven. He imported frozen food — had for decades, paying German suppliers on thirty-day terms the way importers do everywhere, wiring money through Laiki Bank in Nicosia with the thoughtless regularity of a man who has done the same thing hundreds of times.
On the morning of March 28, 2013, when the banks reopened after ten days of closure, he arrived at Laiki needing to send €37,000. The teller informed him that cross-border transfers above €5,000 now required approval from a Restrictive Measures Committee operating out of the Central Bank of Cyprus. Larger transfers required case-by-case review. Cash withdrawals were capped at €300 per day, and you could carry no more than €1,000 out of the country. "We normally have two months' worth of credit from our German suppliers," he told Al Jazeera. "At the moment we don't know if we will continue to enjoy that."
He would not.
What had happened was, by the standards of financial crises, almost surgical. The Eurogroup had offered Cyprus a €10 billion bailout, conditional on depositors absorbing losses. When Parliament rejected a levy on insured deposits, the ECB threatened to cut emergency liquidity entirely. The revised deal wound down Laiki Bank and seized 47.5% of uninsured deposits over €100,000 at Bank of Cyprus. Uninsured depositors at Laiki would eventually recover about six cents on the euro.
The capital controls that followed were the first ever imposed within the eurozone — a monetary union built on the promise that a euro is a euro regardless of where it sits. That promise died in March 2013. Within weeks, London-based firms were purchasing frozen deposit claims at steep discounts. A secondary market in trapped money had opened.
These controls were announced as temporary, which they were after a fashion. Two years later in April 2015 they lifted, by which point Cyprus's economy had contracted 13% and unemployment had hit 17.3%.
The problem was bone deep. The eurozone's payment infrastructure — the most integrated monetary system on earth — rested on a delicate political consensus. When that broke down, the fact that banks were solvent and SWIFT functioned as normal did not matter. Nothing mattered. The promise of the Euro had failed.
The European Union understood the danger of this, or seemed to. In the aftermath, it built the Bank Recovery and Resolution Directive and reformed deposit guarantees, two pillars of a new Banking Union — the most sweeping restructuring of European financial governance since the euro's creation, all designed to ensure that what had happened in Nicosia could not happen again.
It took less than two years to find out that they had not done enough.
Greece, 2015: Four years at €60 per day
On the evening of June 26, 2015, negotiations between Greece and its creditors collapsed.
Prime Minister Tsipras announced a referendum. That night, the ECB froze emergency liquidity assistance at €89 billion, a ceiling that would not rise to accommodate the deposit flight already underway.
Forty-two billion euros had already left Greek banks since November of 2014. On June 28th, capital controls were imposed. Two days later, Greece missed a €1.6 billion payment to the IMF, the first developed country to enter arrears.
The machinery of restriction was familiar from Cyprus, but scaled to a larger economy. ATM withdrawals were capped at €60 per day. A five-member Banking Transaction Approval Committee authorized approximately €20 million in total daily outbound transfers for the entire country, prioritizing medicines and food. Foreign banks began requiring full cash collateral to accept letters of credit from Greek institutions.
Apostolos Kosmidis ran a bird store in Athens — canaries from the Netherlands, bird feeders from Poland, pet food from Austria — and after the controls came down he couldn't pay any of his suppliers. "If we keep going like this," he told NPR, "we're going to close." Kiriakos Kaplanoglou, who sold metal plating materials sourced from Germany and Italy, had the inverse problem: his suppliers had already shipped goods he could no longer pay for, and they would ship nothing more until he could. By September, 58% of companies reported significant disruption. Nearly a fifth had reported a drop in turnover of more than half.
The government, owing pharmaceutical companies more than €1 billion, imposed export bans on 25 drug types to prevent parallel trade. The controls were trapping medicine alongside money.
These controls lasted four years and two months, twice as long as Cyprus. The infection was removed on September 1st, 2019, but only after monumental damage to the Greek people and their trust in the FX system.
What this showed was that despite Europe’s best efforts, treatment that addresses symptoms without altering the architecture that produced the disease only allows it to come back twice as strong. Cyprus and Greece shared a currency, a central bank, a regulatory framework, and now a Banking Union designed to prevent exactly what happened. What they also shared was a structural dependency on political consensus to keep money moving — and when that consensus fractured, no amount of institutional reform made a difference to Apostolos Kosmidis and his canaries.
But political fragility is only one layer of the vulnerabilities that plague the global FX market. The infrastructure has other organs, and other ways to fail.
Afghanistan, 2021: Paying doctors through hawala
On August 15, the Taliban entered Kabul. Within hours, the United States froze Da Afghanistan Bank's reserves: approximately $7 billion at the Federal Reserve Bank of New York, another $2 billion elsewhere. The IMF shut off financing. The World Bank suspended disbursements. Foreign donors who had financed up to 80% of total public expenditure halted funding simultaneously.
Everything went: Correspondent banking, SWIFT access, even Western Union — which suspended services on August 19. The World Bank's Sehatmandi program, which had funded some 2,200 healthcare facilities across the country, shut down overnight.
GDP contracted 20.7% in a single year. Per capita income fell by a third. Seven hundred thousand civil servants went months without pay. Afghans queued for hours to withdraw their own savings from banks that could dispense only $200 a week, a cap that rose in small increments over the next two years, as the emergency calcified into a desperate, new normal.
In a country where 90% of financial transactions had historically moved through hawala, informal value transfer networks rooted in eighth-century Islamic trade, the formal system's collapse didn't create a vacuum so much as reveal how thin the formal layer had always been. The International Committee of the Red Cross launched its Hospital Resilience Project in November 2021, paying roughly 10,000 health workers across 33 hospitals through a patchwork of formal channels and hawala networks. Doctors serving 26 million people were being paid through a mechanism that predates modern banking by twelve centuries.
The most extraordinary workaround was the UN cash airlift. Beginning in December 2021, the United Nations purchased physical US currency from the Federal Reserve Bank of New York and flew it to Kabul. About $40 million per shipment, every ten to fourteen days. By mid-2024, cumulative shipments exceeded $2.9 billion, physical banknotes on cargo planes, flown halfway around the world, distributed through 19 UN entities and 48 international organizations.
This was necessary, because the digital infrastructure that was meant to ameliorate this absurdity had just been shut off.
Meanwhile, $3.5 billion in Afghan central bank reserves sat in a Geneva escrow fund at the Bank for International Settlements, accumulating interest since September 2022. Disbursements as of early 2024: zero.
Cash on cargo planes. That’s an extreme example of what the FX market looks like when all of its vulnerabilities coalesce, but you do not need something that extreme to see this sort of systemic failure.
Nigeria, 2022-2024: $850 million trapped behind a currency wall
₦7.6 million in September 2022. ₦27 million in January 2025. The same degree, the same university, the same tuition in pounds. A 3.6x increase in what a Nigerian family paid, not because the price changed but because the Central Bank of Nigeria maintained, for years, the fiction that the currency hadn't.
You don't need a catastrophe to break the system. You need a central bank defending a peg the market doesn't believe in.
The official naira sat at ₦461 to the dollar. This was a fiction.
Actual transactions occurred in a parallel market where the price hovered between ₦735–750. Nothing had been hacked or frozen. The infrastructure was intact, no sanctions had shut off the financial system from above. The market just lacked buyers willing to pay the price the government was willing to honor.
Airlines felt it first, because their economics made the dysfunction impossible to absorb. Emirates, earning naira ticket revenue it couldn't exchange for dollars, watched its trapped funds climb past $85 million and suspended all Nigerian flights in September 2022. Industry-wide, blocked funds in Nigeria rose from $282 million in April 2022 to $812 million a year later, more than a third of the global total.
The 100,000-plus Nigerian students studying abroad faced arithmetic that worsened by the month. International card payments had been banned. Official FX applications took eight months to process. One widow told Daily Trust she had been taking cooperative loans to pay her son's fees abroad. Parents were pulling children out of foreign universities.
Medical tourism, which had drawn hundreds of Nigerians abroad monthly at a cost of $2.3 million a year, collapsed: by early 2025, FX utilization for medical purposes had fallen 97.4%. The need hadn't diminished, the currency wall had simply made it impossible.
Currency policy dysfunction is endemic, present in dozens of countries at any given moment, activating the same failure modes as a coup or a heist, without the headlines.
Sixteen days after President Tinubu's inauguration in June 2023, the Central Bank unified the exchange rate. The naira fell 36% immediately, eventually sliding to ₦1,400–1,600 per dollar. But the float worked: by April 2024, 98% of blocked airline funds had cleared. The cure required accepting the devaluation the peg had been designed to prevent.
Nigeria's payment crisis ended through a policy correction, not infrastructure reform, not new technology. The rails had been there the entire time. What broke was the policy environment surrounding them, and that environment can break again anywhere a government maintains a fiction about the value of its currency.
Red Sea, 2023-2024: When ships can't move, neither can money
Every failure FX infrastructure has produced up to this point — political, technical, institutional, monetary — lived within the financial and digital layers of the system. They could be understood in the vocabulary of finance.
The Red Sea broke that frame.
Djibouti imports approximately 90% of its food. A small country at the southern entrance to the Red Sea, its ports serve a function far larger than its economy. It is the primary gateway for humanitarian aid to Ethiopia and Somalia, sole export corridor for Ethiopian coffee, the country's largest source of foreign currency.
When Houthi forces began attacking commercial shipping in November 2023, Djibouti's exposure was immediate. Roughly 31% of its foreign trade transits through the Suez Canal. The ships carrying its staples were being rerouted around the Cape of Good Hope, adding 4,500 nautical miles and ten to twenty days to voyages calibrated to the week.
The Canal handles about 12% of global trade. Container transits plunged 41% in the first weeks; by mid-2024, traffic had fallen roughly 90%. Every additional day at sea rippled through a financial architecture built on precise timing.
Letters of credit typically carry validity periods of between 60 and 90 days, with document presentation deadlines of 21 days after shipment. Extend a voyage by three weeks and the documents arrive after the credit expires. The payment will be rejected. Amendments run $50–300 each, and capital frozen for the duration.
A food importer told the [USDA that shipments from Turkey's Port of Mersin to Djibouti, normally 15 to 20 days, now took 55](https://apps.fas.usda.gov/newgainapi/api/Report/DownloadReportByFileName?fileName=Ethiopian+Agricultural+Exports+Thrive+Despite+the+Red+Sea+Shipping+Disruption++_Addis+Ababa_Ethiopia_ET2024-0013.pdf#:~:text=import shipments from Port of Mersin in,transit time has increased to 55 days.). A consignment from Busan, South Korea was rerouted through Ningbo and arrived in 90 days instead of 30. Container freight from Shanghai to Djibouti tripled. War risk premiums, previously nominal at 0.05% of hull value, peaked near 1% — roughly $1 million per voyage for a $100 million vessel.
What this all produced on the ground, was simpler and worse.
Tameru Tadesse runs a coffee export firm out of Ethiopia, shipping to buyers in China and Japan. Before the crisis, at least one vessel arrived in Djibouti daily. By February 2024, it was one a month. "We try not to miss this single vessel," he told The Reporter. "Our coffee is stranded in Djibouti." He explored routing through Mombasa, it was too far away, its roads were too bad.
Mogadishu was closed by diplomatic tensions between Ethiopia and Somalia. Ethiopian coffee supports [five million farming households, fifteen million people in total](https://www.sciencedirect.com/science/article/pii/S2666188825010202#:~:text=There are four coffee production,income [74%2C78].), and moves through a chain built entirely on credit: growers deliver to suppliers on trust, suppliers to exporters, with payment flowing back down after export. When the ships stopped, that chain seized from the top. Exporters who couldn't ship couldn't pay suppliers, who couldn't then pay farmers. "There's no cash in the market now," said Gizat Worku, president of the Ethiopian Coffee Exporters Association.
By mid-2024, the IPC projected that 285,000 people in Djibouti — nearly a quarter of the population — would face acute food insecurity, with 53,000 in emergency conditions. Child malnutrition climbed to 14.7%, up from 13% the year before. In a country where 79% of the population lives in poverty and 90% of food arrives by ship, the disruption became a nutritional emergency.
The word "rails" as it relates to FX has always been a metaphor. The Red Sea stripped it to the metal underneath. Every digital system was functioning, but the goods couldn't move. When the goods can't move, the documentary credit chain seizes, and so does the informal credit chain of growers who delivered their coffee on trust. The infrastructure's fragility ran through a strait 26 kilometers wide, and when that strait became unusable, no financial system could compensate.
As of early 2026, despite months without attacks, Suez traffic remains 60% below 2023 levels. Carriers, once rerouted, have not returned.
Modern Rails: A first but not final step
Thirteen years of intermittent failure left gaps in the architecture, and other infrastructure moved in to fill them.
When Turkey's lira lost 44% of its value in 2021, crypto trading volume on Turkish exchanges exceeded $170 billion.). The surge wasn't speculative enthusiasm, it was a population watching its savings evaporate in real time and reaching for any instrument denominated in something other than a currency shedding value by the week. Much of that volume moved through stablecoins.
In Nigeria, where the parallel market premium made official remittance channels economically irrational, USDT became a de facto transfer rail, not because anyone marketed it as one, but because the formal system had made itself unusable.
A worker in London sending money home through official channels lost 40% or more to the gap between the rate the bank offered and the rate the family needed. Stablecoin transfers, settled peer-to-peer at something close to the parallel rate, routed around the dysfunction entirely.
In Argentina, where the gap between official and parallel exchange rates has routinely exceeded 100%, dollar-pegged stablecoins have become savings technology for the middle class, a way to hold purchasing power outside a banking system that has failed to preserve it through multiple currency crises across multiple decades.
Stablecoins respond to specific failures in the traditional FX architecture: thinning correspondent banking chains (down 22% in a decade), multi-day settlement windows during which a currency can lose a third of its value, branches closing on weekends, and policy dysfunction that makes official channels irrational.
But the specificity cuts both ways. Stablecoins can't move Tadesse's coffee out of Djibouti or fix the regulatory vulnerability Cyprus exposed. A government that wants to block on-ramps can do so, as China has demonstrated. And Stablecoins carry their own fragility: Issuer solvency, regulatory uncertainty, reserve opacity. TerraUSD collapsed in May 2022, erasing $40 billion in days. The asset class is not exempt from the risks it claims to address.
Even so, the pattern of adoption tracks the failures almost exactly. Stablecoin volume grows where correspondent banking has retreated, informal corridors open where formal channels have priced themselves out of reach. This is tissue forming around damage, not a solution to all the body's ills.
The view from right now
What this survey reveals is not a broken system but a brittle one, infrastructure that works until it doesn't, with the failure modes concentrated on those who can least afford them.
A multinational corporation facing Greek capital controls could restructure through Cypriot subsidiaries, or hire lawyers to navigate approval committees, but a small importer watching his canary shipment rot for lack of payment can’t.
An international bank facing compliance requirements can absorb the cost, but a regional bank in the Pacific Islands – where correspondent relationships have declined 50% since 2011 – may find itself cut off entirely.
Remittances totaled $656 billion to low- and middle-income countries in 2022—more than foreign direct investment and official development assistance combined. The infrastructure that carries them was not designed for resilience. It was designed for a world where currencies remained stable, correspondent banks remained willing, and ships passed freely through chokepoints.
That world exists “most of the time.” But “most of the time” is looking more and more rare these days.
The UAE is home to 8.7 million migrant workers, 93% of the country's population. Last year they sent nearly $50 billion home – half to India, Pakistan, and the Philippines – much of it through exchange houses because they earn below the minimum to open a bank account.
As of the week of March 1st, the airspace above the Gulf is closed. Airports in Dubai and Abu Dhabi have been struck, more than 11,000 flights have been canceled. The Philippines has imposed a deployment ban and India is running evacuation flights. Fifty-eight thousand Indonesian workers are stranded in Saudi Arabia during Ramadan.
A crisis is growing across both humanitarian and economic fronts, one that will once again put the vulnerable FX market to the test.
Frequently Asked Questions
How does correspondent banking actually work, and why does it create fragility?
Most banks don't have direct relationships with banks in other countries. Instead, they route payments through intermediaries that hold accounts on their behalf. A remittance from the Philippines to Nigeria might pass through three or four institutions: the sender's bank, a regional correspondent, a global clearing bank in New York or London, another correspondent, and finally the recipient's bank. Each handoff accrues time and costs, and requires both parties to trust each other's compliance standards. When a correspondent bank decides a country or institution is too risky due to sanctions exposure or weak anti-money-laundering controls, for example, it "de-risks" by ending the relationship. The affected bank often loses its only pathway to entire regions. This is why correspondent banking relationships have declined 22% globally since 2011, with the steepest drops in the Caribbean and Pacific Islands.
What happens to money mid-transit when a currency suddenly devalues?
This is called settlement risk, and it's one of the least understood fragilities in cross-border payments. Most international transfers don't settle instantly, they can take five days or more depending on the currencies involved, and the number of intermediaries. During that window, the money exists in a kind of limbo. If the destination currency devalues sharply while the transfer is in flight, the recipient gets less value than the sender intended. Egypt's March 2024 devaluation is a stark example: remittances that entered the system before the float and settled after it lost over a third of their value in transit. The sender did nothing wrong; the infrastructure simply couldn't move fast enough to outrun the policy change.
How common are major disruptions in the FX market? Isn't this essay cherry-picking extreme cases?
The cases in this essay are extreme by design, they're the moments when the market's vulnerabilities were the most pronounced and diagnosable. But they're not outliers in the statistical sense. The World Bank tracks remittance costs and corridor functionality across 365 country pairs. At any given time, dozens of corridors are degraded: fees above 10%, multi-day delays, informal channels carrying more volume than formal ones. Currency crises severe enough to distort remittance flows occur somewhere in the world almost constantly: Argentina, Turkey, Egypt, Nigeria, Pakistan, Sri Lanka, Lebanon, all within the past five years. This essay seeks to document an established pattern of chronic conditions punctuated by acute crises, with the acute crises acting as illustration.
Why do capital controls last so much longer than governments initially claim?
Governments impose capital controls in emergencies, promising they'll be temporary. But unwinding them requires solving the problem that triggered them, and that problem is usually a loss of confidence. Cyprus couldn't lift controls until depositors stopped fleeing; depositors wouldn't stop fleeing while controls signaled the system was unstable. Greece couldn't lift controls until the banking system was recapitalized; recapitalization required foreign investment that wouldn't arrive while controls were in place. The controls create a kind of credibility trap: their existence is evidence that what they sought to ameliorate is still there, which makes ending them harder. The political incentive is always to announce controls as temporary, but unfortunately, crises of confidence don't commonly operate on a predictable schedule.
What would resilient cross-border payment infrastructure actually look like?
Resilience in this context means three things: reduced concentration (no single chokepoint), reduced settlement time (less exposure to volatility during transit), and reduced dependency on discretionary gatekeepers (fewer points where a single institution's risk appetite determines access). Some of this is already being built: the BIS's Project Nexus aims to link fast domestic rails directly, targeting 60-second settlement. Some countries are experimenting with central bank digital currencies that could eventually enable direct central-bank-to-central-bank transfers. Stablecoins offer a partial workaround for the correspondent banking problem in some corridors, but resilience isn't a technology, it's a design principle. A system with faster rails but the same concentration points just fails faster. The hard problem is architectural.
Why are migrant workers disproportionately affected when payment infrastructure fails?
Three structural reasons: First, they lack alternatives. A multinational corporation facing capital controls can restructure through foreign subsidiaries, or hire lawyers to navigate regulatory slowdown. A construction worker in Dubai sending $400 to Kerala has one channel, maybe two. When that channel fails, there's no backup. Second, they're often paid in the currency under stress and need to convert it quickly to send home, meaning they absorb FX losses that corporations can hedge against. Third, they frequently fall below thresholds that would give them access to more resilient systems. The infrastructure wasn't designed to fail on them specifically, it wasn't designed for them at all.
Share article
Read other articles
Stay informed with our latest articles on currency launches, institutional FX trends, and global liquidity.





