The Triffin Dilemma: Why the Reserve Trap Still Binds

Triffin diagnosed the reserve currency trap in 1960. The gold window closed in 1971 and was supposed to kill it. The trap survived in different clothes, and the 2026 attempt to reshore manufacturing is running straight into the same wall.

Marty Bent

August 21, 202618 min read

Empty assembly floor of an abandoned American manufacturing plant, idle crane rails overhead and bare concrete where machinery was removed

Robert Triffin walked into Congress in December 1960 and told the Joint Economic Committee that the dollar system had a fatal contradiction baked into it. He wasn't wrong. He was just early, and he got the direction of the failure partly backward. That's the interesting part, and it's the part that makes the Triffin dilemma worth taking seriously even after the specific mechanism he described (gold convertibility) died on August 15, 1971, when Nixon closed the window.

The argument I'm going to make here is that the shape of the trap survived 1971 even though the gold constraint didn't, that the binding element just changed form, and that the 2026 effort to reshore American manufacturing is running straight into the same structural wall Triffin identified. I'll also give you the strongest case that I'm wrong, because Bordo and McCauley made it well enough that ducking it would be intellectually dishonest. Then I'll tell you why I think their objections hit the 1960 version of Triffin harder than they hit the version that actually operates today.

What the Triffin Dilemma Actually Says

Start with the mechanism as Triffin stated it, precisely, because most of what passes for "Triffin dilemma" commentary online skips this and goes straight to the conclusions.

Triffin published two articles in March and June of 1959, then assembled the argument in full in Gold and the Dollar Crisis: The Future of Convertibility (Yale University Press, 1960). The testimony before the Joint Economic Committee in December 1960 is the capsule version. Under the Bretton Woods system, with the dollar convertible to gold at $35 an ounce, the mechanism ran as follows:

The world needs dollars for reserves and trade settlement. Dollars reach the world only via US balance-of-payments deficits. Those deficits accumulate as foreign dollar claims on the United States. Eventually foreign claims exceed the US gold stock. Confidence breaks, a run on gold follows, the dollar supply contracts, and the result is global deflation.

That is the four-step sequence, and every word of it is specific to gold convertibility. There is a run because there is something to run on. The trigger is a hard asset held in finite quantity. When foreign claims against that asset exceed the asset itself, the system is technically insolvent, and runs on insolvent institutions are what rational creditors do.

Nixon closed the gold window on August 15, 1971. That was the run Triffin predicted, arriving. Nixon preempted it by ending convertibility before the run could complete. The system didn't deflate; it inflated. And that difference is where the whole modern debate lives.

How the System Broke, and Why the Dilemma Didn't Die With It

The gold-convertibility mechanism is dead. There is no gold window to run on. So anyone who reads Triffin's 1960 book and concludes the dilemma is still operating in exactly its original form is wrong, and Bordo and McCauley are right to call that out.

But the question worth asking is whether the underlying structural pressure (that reserve-currency issuance creates forces incompatible with the issuer's domestic economic stability) survived the death of gold convertibility. I think it did, and the post-1971 reformulation is what the ECB's Lorenzo Bini Smaghi described in his 2011 lecture The Triffin Dilemma Revisited: the binding constraint shifts from gold drain to structural currency overvaluation.

Here is the transition stated plainly, because articles that paper it over are the ones that deserve the Bordo and McCauley takedown. Under Bretton Woods, the cost of reserve status was gold. The US had to keep draining its gold reserves to supply the world with dollars, until the claims exceeded the metal and confidence broke. After 1971, with no gold convertibility, there is no run. But the reserve demand doesn't disappear. The world still needs dollars. And that constant, structural extra demand for the dollar, from trade invoicing, from reserve accumulation, from offshore dollar borrowing, keeps the dollar priced higher than it would otherwise be. A persistently overvalued currency means that producing lower-margin tradable goods domestically is uncompetitive. Not quarter to quarter, not visibly as a single event. Slowly, over decades, the factory floor migrates.

The way I'd put it after our May 2025 conversation with Lyn Alden: under Bretton Woods the cost of reserve status was draining gold, and today it is hollowing out the industrial base. Alden described the long-run consequence directly. "We keep sending out little parts of our industrial base over time," she said, "to maintain the global reserve currency status."

Same shape, different binding constraint. That is the modern reformulation, and it is the version I'll defend.

The Strongest Case That It's a Myth

Bordo and McCauley's paper, "Triffin: dilemma or myth?" (NBER Working Paper 24195, also published as BIS Working Paper 684), is the most rigorous attack on the Triffin framework, and it deserves a straight answer rather than a rhetorical sidestep. Their case has three distinct blades.

Blade one: Triffin got the direction wrong. He forecast that the United States would behave prudently and that the world would suffer from a deflation of dollar supply as foreign claims overwhelmed the gold stock. What actually happened was US fiscal profligacy and global inflation, the structural opposite. If the prediction misfired that badly, the analytical framework deserves scrutiny, not reverence. This is not a minor objection. Triffin's specific scenario didn't just arrive late; it arrived in the wrong direction.

Blade two: the historical comparator didn't collapse. Bordo and McCauley note that the US gold position in the postwar period was no worse, relative to foreign claims, than the UK's position around 1900. The UK ran the global reserve system from that position for another two decades without a terminal run on sterling. So the mechanism Triffin described wasn't as mechanically inevitable as he suggested. Systems can run above their theoretical breaking point for longer than models predict.

Blade three: the modern "fiscal Triffin" is overbuilt. The contemporary restatement (that global safe-asset demand forces the US to issue excessive debt) overstates both the scale of that demand and the inflexibility of its supply. Other sovereigns can and do issue safe assets. The US is not a monopoly supplier. The demand is not perfectly inelastic. Their prescription is that the US should concentrate on maintaining the credibility of its willingness to pay, not on resolving a structural dilemma that may be more myth than mechanism.

That is the case at full strength. Now here is my answer.

Why I Think the Mechanism Still Binds

On blade one. Bordo and McCauley are right that Triffin's specific prediction misfired. But a prediction failure is not the same as a mechanism failure. Triffin identified a real structural pressure, that reserve-currency issuance is incompatible with domestic economic stability, and got the direction of the terminal instability wrong because Nixon preempted the gold run by closing the window before it completed. The system didn't deflate; it inflated instead. The constraint changed form, not existence. Profligacy and inflation are what you get when you remove the gold brake and keep running deficits anyway. That's a different flavor of instability from the one Triffin predicted, but it's still instability, and it's still downstream of the same structural logic.

On blade two. The UK-in-1900 comparison is the strongest blade, but it is specific to the gold era. The UK's position was constrained by exactly the same gold-convertibility mechanism Triffin described. The fact that it didn't immediately collapse doesn't mean the mechanism was wrong, it means systems can run above their theoretical breaking point longer than models suggest. And the sterling system did eventually collapse; it just took fifty more years and two world wars. More importantly, the comparison applies to the 1960 Triffin, not the post-1971 version. The modern overvaluation channel has no run-on-gold event as a trigger. The pressure bleeds out slowly, as manufacturing capacity migrates and the political economy fractures. Bordo and McCauley's strongest objection lands squarely on the version of Triffin that already ended. It doesn't land as cleanly on the version operating now.

On blade three. Bordo and McCauley are right that the US is not the monopoly supplier of safe assets and that supply has some flexibility. But the overvaluation channel doesn't require monopoly or perfect inelasticity. It requires that reserve demand structurally bids the dollar above its trade-equilibrium level, and that is not a theoretical claim. It is documented in the manufacturing employment data.

EPI's 2015 analysis by Robert Scott attributed 72% of the five million manufacturing jobs lost between 2000 and 2014 to the growing manufacturing trade deficit, not to productivity growth. Scott specifically named currency overvaluation as the leading cause, noting that the dollar rose 19.7% against major currencies from December 2013 to March 2015. To be fair, CEPR and EIG attribute more of the job loss to productivity and automation, that causation dispute is real and I'm not going to hide it. But the EPI analysis is mainstream empirical work, not Bitcoin advocacy, and the overvaluation channel it documents is consistent with the mechanism Triffin described in its post-1971 form.

Bordo and McCauley's prescription (focus on credibility of willingness to pay) is sensible as far as it goes. But it addresses the fiscal version of the dilemma while largely ignoring the industrial version. Both exist.

I'll also make the concession Bordo and McCauley are entitled to: the term "Triffin dilemma" has been stretched to cover several distinct phenomena (gold drain, fiscal expansion, safe-asset monopoly, currency overvaluation) that don't all operate through the same mechanism. The label-stretching invites the confusion they're trying to correct. When I use "Triffin dilemma" in this piece, I mean specifically the overvaluation-and-deindustrialization channel, which is the version I think binds today. I try to be precise about when I'm talking about the 1960 gold mechanism versus the post-1971 reformulation.

In our May 2025 conversation, Alden addressed the implicit Bordo and McCauley question of why the system hasn't broken yet: "There's literally way more inflexible demand for dollars than there are dollars in the system. And that's why this system kind of perpetuates itself for such a long period of time. It's not as simple as a bunch of countries getting together and deciding to repudiate the currency." That's why the dilemma persists without triggering an immediate collapse. The demand is structural, not sentimental. By her framing the system runs at roughly 20-to-1 leverage, something like $5.8 trillion of base money against more than $120 trillion of dollar-denominated debt held at home and offshore, and you do not vote your way out of a short position that size.

The Recycling Loop and the Rust Belt

The overvaluation channel doesn't operate in the abstract. It runs through a specific mechanism: the dollar surplus recycling loop. Trace it and the human cost becomes impossible to miss.

The world needs dollars. The only way the world gets dollars in quantity is through US trade deficits, hundreds of billions, lately over a trillion dollars per year in net outflows. The other side of that deficit is a capital-account surplus: the dollars the world earns get sent right back as inflows into American stocks, bonds, real estate, and Treasuries. Viewed honestly, the foreign sector is an intermediary in a domestic transfer. In our May 2025 conversation, Alden described the endpoint directly: the system is "constantly taking economic vibrancy out of Michigan and Ohio and rural Pennsylvania" and directing it into financial assets on the coasts.

The BLS data makes the Rust Belt story concrete. Manufacturing employment peaked in June 1979 at 19.6 million workers. By June 2019 it had fallen to 12.8 million, a loss of 6.7 million jobs, a 35% decline. The share of total nonfarm employment tells the story even more starkly: manufacturing went from 22% of all US jobs in 1979 to 9% in 2019. Apparel and textiles lost 81% of their workforce. Employment never fully recovered after any of the five recessions that followed 1979. Those are Katelynn Harris's figures in Forty Years of Falling Manufacturing Employment, published November 2020, and they predate the COVID disruptions entirely.

The causation question is genuinely contested, and I'm not going to hide that. CEPR and EIG researchers attribute more of the manufacturing decline to productivity growth and automation than to trade. That view has serious economists behind it. EPI's Scott puts 72% of the 2000-2014 job loss on the growing manufacturing trade deficit; the counter-view says a lot of those jobs would have been automated regardless of the trade balance. Both things can be partly true simultaneously.

What the dispute cannot do is dissolve the overvaluation channel entirely. Even if automation explains some of the job loss, currency overvaluation that makes domestic production uncompetitive is a documented, additional pressure operating through the mechanism Triffin described.

In our January 2025 conversation, Nik Bhatia described the other side of the recycling loop from his vantage point inside the eurodollar plumbing: repo-financed demand for financial instruments as a concrete example of how dollar recycling flows into US financial assets rather than into productive domestic capacity. The dollars come back in, they inflate financial asset prices, and the people who own financial assets, disproportionately on the coasts, capture the gains while the manufacturing base continues its slow erosion.

2026: The Trap in New Clothes

The most useful test of any economic framework is whether it explains current events. By that standard, the post-1971 Triffin mechanism is doing well.

Stephen Miran's November 2024 paper, A User's Guide to Restructuring the Global Trading System (Hudson Bay Capital), names Triffin's dilemma directly and accepts that the dollar is overvalued by its reserve role. It lays out a menu of responses: mild tariffs as a stick, a negotiated currency accord as a carrot, terming out the debt by convincing foreign holders to extend duration, and explicitly elevating gold and cryptocurrencies as neutral reserve assets to absorb flows that would otherwise pile into US assets. This is a serious document from a person who became chair of the Council of Economic Advisers, not a fringe argument.

In our May 2025 conversation, Alden gave the Miran framework a split assessment: sound diagnosis, messy execution. "Tariffs went up super high, super quickly" while reshoring a manufacturing base takes years. The pain landed immediately; the benefit didn't. The plan also overestimates the US negotiating position, the map of which country is China's largest trading partner has shifted dramatically over the past two decades, and that changes the bargaining calculation.

The reserve data tells its own story, and it requires careful reading. The IMF's COFER data for Q1 2026, published June 30, shows the dollar's share of allocated reserves at 57.13%, up from 56.42% in Q4 2025. That is a quarterly rise, and anyone running a "dollar collapsing now" narrative has to reckon with it. About half that rise is FX valuation effect, not active buying, but the number is the number.

The long arc is still down from over 70% in 2000, and the quarterly tick-up contradicts the doom framing without dissolving the structural trend.

The single most striking fact in the 2026 reserve data is the IMF's statement that in 2025, gold surpassed US Treasuries as a share of official reserves. The same data brief notes this was "driven almost entirely by gold price valuation effects," not a Treasury fire sale. Both facts matter. The gold crossover is real; the mechanism behind it is mostly price appreciation, not active dumping of Treasuries.

In our June 2026 conversation, Luke Gromen cited ECB figures showing gold at 27% of global central-bank reserves against 22% for Treasuries, and framed it as evidence that the post-1971 reserve architecture is shifting. His figure and the IMF's measure slightly different things, official reserves overall versus the allocated FX reserves COFER tracks, which is why the percentages don't line up.

On the renminbi: it sits at 1.99% of global reserves, up from 1.95% the prior quarter. Any "the yuan is replacing the dollar" framing dies on that number. China's yuan-gold settlement plumbing is real. Gromen described in our June 2026 conversation how offshore yuan clearing operates in major gold hubs and how Swiss gold export flows toward the Gulf suggest some energy is being settled outside the dollar.

The architecture of an alternative is being built around the margins. Its scale in the reserve data is rounding error. Both things are true simultaneously.

The current account deficit narrowed through 2025 to $1.12 trillion, or 3.6% of GDP, down from 4.0% in 2024, a 5.8% year-over-year improvement. Q4 2025 came in at 2.4% of GDP. Q1 2026 widened back to 2.9%. A "nothing is being fixed" framing has to acknowledge the 2025 narrowing.

One year of narrowing doesn't dissolve a forty-year structural dynamic, and the Q1 2026 widening suggests the improvement may not be durable. But the narrowing happened, and intellectual honesty requires saying so.

In our July 2026 conversation, Michael Howell pointed to a different indicator of fiscal constraint becoming binding: nominal GDP running at 6-7% against a suppressed 10-year yield, a gap that, in his chart going back to 1955, always closes by yields rising rather than by the economy slowing. He also noted that 80% of gross US issuance is now under two years in maturity, evidence of ongoing monetization through the bill channel, which connects the fiscal side of the dilemma to the reserve-demand side.

The bond market is becoming the binding constraint that gold once was. In our June 2026 conversation, Gromen walked the fiscal arithmetic directly: interest plus entitlements already consume nearly all federal receipts, and the deficit math has no clean exit. The profligacy Bordo and McCauley correctly identified as the actual outcome of the post-Bretton Woods system has compounded to the point where the system is approaching a different kind of constraint, not a gold run, but a bond-market reckoning.

What Comes After the Reserve Asset

The honest answer is that nobody knows the timeline, and I'm not going to pretend otherwise. Reserve currencies don't die quickly. In our May 2025 conversation, Alden made the point that the inflexible demand for dollars is the precise reason near-term dollar-doom calls keep being wrong. The network effects are real. The exit, if it comes, is gradual, not sudden.

What the reserve data does show is where officialdom is already moving. Central-bank gold tonnage bottomed in 2009 and has climbed for fifteen years, accelerating after Russia's reserves were frozen in 2022, with 2022 itself setting a record 1,136 tonnes of net central-bank buying. That acceleration is rational behavior: if your reserves can be weaponized, you move toward an asset that can't be frozen by anyone. Gold is that asset. It has no issuer to sanction.

In our June 2026 conversation, Gromen put the logic plainly: "There's nothing more bullish for a neutral reserve asset than sovereign insolvency." Gold leads because it's where officialdom reaches first. The plumbing for settling trade in gold is already being built. Gromen described how offshore yuan clears in major gold hubs and how some energy appears to be settling outside the dollar, with the token used for the conversion mattering less than the gold changing hands. His line on the currency question: "It's which Chuck E. Cheese token you use to buy the gold. Doesn't really matter."

Bitcoin is a different case, and I'll be direct about where it stands. It's a smaller asset than gold and more volatile, which keeps it out of official reserve discussions for now. But it has a structural feature gold doesn't: it's an open, permissionless, non-dollar settlement network. As Alden put it in our May 2025 conversation, once Bitcoin reaches a scale where volatility has structurally declined, "that's in the discussion," not because governments decide to put it there, but because it will be large enough and liquid enough that the discussion becomes unavoidable.

It starts with people before it starts with governments, which is why it lags gold in the official-reserve data even as the same insolvency forces push in its direction.

In our January 2025 conversation, Bhatia described one specific mechanism: sovereign bonds issued to build Bitcoin reserves, financed through London eurodollar repo markets, the same offshore plumbing that has funded every major emerging-market borrowing cycle. The mechanism by which reserve-asset demand for Bitcoin could build is the same plumbing that built demand for every prior reserve asset. It doesn't require a revolution. It requires repo desks deciding the collateral is good.

The Triffin dilemma has no clean resolution under the current system. A reserve currency with no central issuer is the only structure the dilemma can't apply to. There's no issuer whose domestic needs conflict with the world's reserve needs, because there's no issuer. That's the structural case for a neutral reserve asset, and it's why I think the exit, when it comes, runs through gold first and Bitcoin eventually, not through another nation-state currency that would inherit the same trap in new clothes.

Frequently Asked Questions

What is the Triffin dilemma in simple terms?

The reserve-currency issuer has to run balance-of-payments deficits to supply the world with its currency. But those deficits eventually undermine confidence in the currency itself. Under Bretton Woods, the confidence break would have been a run on gold; Triffin predicted it in 1960 and Nixon preempted it in 1971 by ending gold convertibility. After 1971, the same structural pressure shifted form: instead of draining gold, the US pays with persistent currency overvaluation that erodes its manufacturing base. Same trap, different binding constraint.

Did the Triffin dilemma end when Nixon closed the gold window in 1971?

The gold-convertibility mechanism ended. The overvaluation channel didn't. Without gold convertibility there's no run to trigger, so Triffin's specific 1960 prediction (dollar scarcity, global deflation) stopped applying. But the post-1971 reformulation replaces gold drain with a different cost: reserve demand keeps the dollar structurally overvalued, which makes domestic manufacturing of lower-margin goods uncompetitive over time. The ECB's Bini Smaghi laid this out in a 2011 lecture on the dilemma revisited. The shape of the trap survived; the specific mechanism changed.

What is the difference between the original Triffin dilemma and the modern "fiscal Triffin"?

The 1960 version was about gold: foreign claims on a gold-convertible dollar would eventually exceed the US gold stock, triggering a run. The post-1971 overvaluation version is about the industrial base: reserve demand keeps the dollar priced above its trade-equilibrium level, making domestic manufacturing uncompetitive over decades. The "fiscal Triffin" is a third framing: that global safe-asset demand forces the US to issue excessive debt. Bordo and McCauley's NBER Working Paper 24195 argues the fiscal version overstates both the demand for safe assets and the inflexibility of their supply, and they have a point on that version specifically. The overvaluation-and-deindustrialization version is harder to dismiss with the same objection.

Is the Triffin dilemma a myth? What do critics say?

Bordo and McCauley's paper is the strongest attack. Their case has three parts: Triffin predicted US prudence and global deflation but got US profligacy and global inflation; the UK ran a similar reserve position in 1900 without a terminal collapse; and the modern fiscal version overstates both safe-asset demand and supply inflexibility. These are serious objections. The honest response is that they land most squarely on the 1960 gold mechanism and the fiscal Triffin, less cleanly on the post-1971 overvaluation channel. The prediction misfired; the underlying structural pressure is documented in the manufacturing employment and trade-deficit data.

How does the Triffin dilemma explain US manufacturing decline?

The mechanism runs through currency overvaluation. Reserve demand keeps the dollar priced higher than it would otherwise be. An overvalued dollar makes lower-margin domestic manufacturing uncompetitive relative to imports. Over decades, production migrates. BLS data shows US manufacturing employment fell from 19.6 million in June 1979 to 12.8 million by June 2019, a 35% decline, with the share of total nonfarm employment dropping from 22% to 9% over that period. EPI's 2015 analysis attributed 72% of the 2000-2014 manufacturing job loss to the growing manufacturing trade deficit, and named currency overvaluation as the leading cause. CEPR and EIG researchers attribute more to productivity and automation, that causation dispute is real, but it doesn't dissolve the overvaluation channel entirely.

What does the Triffin dilemma mean for the dollar today?

The IMF's COFER data for Q1 2026 shows the dollar's reserve share at 57.13%, up from 56.42% the prior quarter, with about half of that rise attributable to FX valuation rather than active buying. The long arc is still down from over 70% in 2000. Gold surpassed US Treasuries as a share of official reserves in 2025, though the IMF notes this was driven almost entirely by gold price appreciation rather than active Treasury selling. The renminbi sits at 1.99% of reserves. The 2025 current account deficit narrowed to 3.6% of GDP from 4.0% in 2024, though Q1 2026 widened back to 2.9%. The structural dynamic is intact; the quarterly moves don't resolve it, and the 2026 reshoring effort is running into the same overvaluation wall Triffin described.

Could gold or Bitcoin replace the dollar as a reserve asset because of the Triffin dilemma?

The Triffin dilemma applies to any nation-state reserve currency, because any issuer has domestic interests that eventually conflict with the world's reserve needs. A neutral reserve asset with no central issuer is structurally immune to the dilemma, because there's no issuer whose balance sheet is the world's reserve pool. Gold is where central banks are already moving. Tonnage has climbed for fifteen years and accelerated after Russia's reserves were frozen in 2022. As Gromen put it in our June 2026 conversation, gold leads because it's where officialdom reaches first, and "there's nothing more bullish for a neutral reserve asset than sovereign insolvency." Bitcoin is smaller and more volatile, which keeps it out of official reserve discussions for now, but it has a feature gold doesn't: an open, permissionless settlement network that doesn't route through any single country's banking system.

What is the "exorbitant privilege" and how does it relate to the Triffin dilemma?

Valéry Giscard d'Estaing coined the phrase in the 1960s to describe the advantage the US gets from issuing the world's reserve currency: cheap imports, cheap foreign borrowing, and the ability to run deficits in currency only the US can print. The Triffin dilemma is the other side of that coin. The privilege and the trap are the same mechanism viewed from different angles. The privilege is that global demand for dollars lets the US consume more than it produces. The trap is that satisfying that demand requires running deficits that keep the dollar overvalued and erode the industrial base, the same deficits that make the privilege possible in the first place. You can't have one without the other.