De-Dollarization Looks Like a Reserve Story. The Settlement Plumbing Says It Isn't.

A single number does most of the work in the de-dollarization debate. The dollar's share of allocated official foreign exchange reserves fell to 56.42% in the fourth quarter of 2025, the lowest reading since the mid-1990s. The story writes itself from there: a slow, orderly exit, decade by decade, with an obvious endpoint.

Then the same series moved the other way. IMF COFER data put the dollar at 57.13% of allocated reserves in the first quarter of 2026, on a total allocated pool of $13.10 trillion, and the IMF attributed roughly half of that increase to valuation effects rather than to anyone buying anything. The euro sat at 20.03%, the yen at 5.44%, and the renminbi at 1.99%.

That is the problem with the reserve-share framing. It is a portfolio statistic being asked to describe a plumbing system. Reserve composition looks like the story of dollar dominance. It isn't. It is the one layer of the system that a central bank can change by voting on it.

First, a Correction to the Standard Origin Story

Most explanations of dollar dominance rest on two load-bearing claims, and one of them does not survive contact with the record.

The durable claim is the Bretton Woods one. The 1944 conference fixed other currencies to the dollar and the dollar to gold at a moment when the United States held the overwhelming majority of the world's monetary gold — an accident of war finance rather than a verdict on economic quality. The arrangement ended in August 1971 when convertibility was suspended. Dollar dominance outlived the system that created it, which is itself the interesting fact.

The claim that does not hold is the petrodollar treaty. The widely repeated version says that in 1974 the United States and Saudi Arabia signed a binding agreement requiring oil to be priced exclusively in dollars, and that this agreement had a 50-year term. There was no such agreement. What existed was a Joint Commission on Economic Cooperation established in 1974, which encouraged Saudi reinvestment of oil proceeds into US goods, services and assets. Historians of the period state plainly that Saudi oil was never contractually required to be sold in dollars. Dollar invoicing became standard because the Saudi oil industry was built and run by US firms that dealt in dollars, and the convention persisted after nationalisation because switching had no payoff for either side.

This is more than a footnote. If dominance rested on a treaty, it would have an expiry date and a counterparty who could walk away. Resting instead on convention and on the size of the funding market, it has neither. The failure mode differs, and so does the timeline.

Four Measures That Refuse to Agree

Across the four standard measures of international currency use, the dollar's share ranges from roughly half to nearly nine-tenths depending on which is chosen.

Layer	Dollar share	Source and date	Nearest rival
Allocated FX reserves	57.13%	IMF COFER, 2026 Q1	Euro, 20.03%
Cross-border payments by value	50.10%	Swift, June 2026	Euro, 21.88%
Trade finance by value	81.16%	Swift, June 2026	Renminbi, 8.00%
One side of FX turnover	89.2%	BIS Triennial, April 2025	Euro, 28.9%

The BIS number gets least attention and carries most weight. Global over-the-counter FX turnover reached $9.6 trillion per day in April 2025, and the dollar was on one side of 89.2% of it, up from 88.4% three years earlier. The euro was on 28.9%, down from 30.6%; the renminbi on 8.5%, ranked fifth. Since every trade has two sides these shares sum to 200%, and the dollar's near-90% reading means it functions as the vehicle currency: a Thai importer paying a Brazilian supplier is likely to route through dollars, because the direct pair is thinner and costlier.

The second BIS figure is more revealing. FX swaps were the largest instrument at $4 trillion of daily turnover, 42% of the total. An FX swap is not a currency bet; it is short-term secured borrowing. That $4 trillion a day is the world rolling its dollar funding, mostly at tenors under a week — the mechanism by which a dollar shortage anywhere becomes a dollar shortage everywhere within hours.

Region by Region: Who Is Actually Diversifying

Euro Area — Real Progress, Wrong Layer

The euro had a genuinely strong year. The ECB's June 2026 assessment put the euro's composite share across international use indicators at around 20% for 2025, with international debt issuance in euro rising roughly 30% against 2024 to its highest level since the currency launched, and the euro becoming the leading currency in green and sustainable international bond issuance for the first time.

None of that has moved the reserve share. The binding constraint is the absence of a single deep safe asset. US marketable Treasury debt outstanding stood at $31.5 trillion in July 2026, up 8.6% year on year, trading at an average daily volume of $1,209.1 billion, up 12.1%. A reserve manager needing to move $20 billion in an afternoon can do it there without moving the price much. The euro area sovereign market is split across issuers with different credit and different liquidity, and joint issuance remains a fraction of the size. Verdict: a winner on issuance, still constrained on reserves.

China — Winning the Invoicing Fight, Not the Funding One

China's progress is real and concentrated exactly where theory predicts. The renminbi is second in trade finance at 8.00% of value, ahead of the euro's 5.59%, because China can require its own counterparties to invoice in renminbi. It is fifth in general payments at 3.10% and holds 1.99% of allocated reserves.

The rails tell the same story. CIPS settled ¥180.15 trillion across 8,441,897 transactions in 2025, with 1,766 participants in 124 countries, up from ¥175.49 trillion and 1,629 participants in 2024. Growth is steady; scale is not yet comparable. The Fedwire Funds Service moved an average of about $4.593 trillion per day across roughly 869,187 daily transfers in 2025 — at prevailing exchange rates, under six business days to clear what CIPS clears in a year. Verdict: winner on bilateral invoicing, not on reserve or funding status.

Japan — The Largest Creditor, and the Most Exposed

Japan held $1,185.5 billion of US Treasury securities at end-December 2025, the largest foreign position, ahead of the United Kingdom at $863.1 billion, mainland China at $684.4 billion, Belgium at $477.3 billion and Canada at $468.3 billion. Total foreign holdings were $9,269.5 billion.

Japanese institutions are also among the heaviest users of FX swaps to fund those dollar assets out of a yen liability base. That makes Japan a beneficiary in normal conditions and a sensitive point in stressed ones: when dollar funding tightens, the yen cross-currency basis widens early and hedged returns on foreign bonds compress fastest. Verdict: creditor on paper, funding-dependent in practice.

Emerging Market Borrowers — The Clearest Losers

This is the layer the reserve-share debate ignores entirely. BIS global liquidity indicators put dollar credit to non-bank borrowers outside the United States at $14.7 trillion at end-March 2026, growing 7.3% over the year, with emerging and developing economies at roughly 30% of the stock — on the order of $4.4 trillion of dollar liabilities held by entities whose revenues are largely not in dollars.

A country can diversify its reserves in a quarter. It cannot re-denominate its firms' existing debt at all. A borrower with dollar liabilities and local-currency revenue faces a balance sheet that tightens when the dollar strengthens, regardless of what its central bank holds. Verdict: structurally short the dollar, with no available substitution.

Official Gold Buyers — Hedging the Right Thing, Slowly

Central bank gold accumulation has continued. Official reserves rose by a net 41 tonnes in May 2026 alone, and 2026 purchases through mid-year were led by Poland at 64 tonnes, Uzbekistan at 33 tonnes, mainland China at 25 tonnes and Kazakhstan at 20 tonnes. The ECB has noted the trend explicitly in the context of persistent geopolitical tension.

Gold answers one specific risk: reserve claims held in another jurisdiction can be immobilised. It does not answer the settlement question. Gold does not pay a supplier, roll a swap or service a dollar bond. Verdict: a rational Layer 1 hedge that leaves Layers 2 through 4 untouched.

Where This Reading Would Fail

Ranking these outcomes gives a consistent pattern: diversification is fastest where the decision is unilateral and the asset fungible, slowest where it requires a counterparty to move at the same time.

• Unilateral and fast. Reserve composition and gold holdings. One institution decides. Observable within quarters.
• Bilateral and medium. Trade invoicing. Both sides must accept the same unit, which is why China's gains are concentrated in trade it can condition.
• Multilateral and slow. Settlement rails, which need network scale to be worth joining, and are not worth joining until they have it.
• Effectively fixed. The existing stock of dollar liabilities. Contracts do not renegotiate on ideological grounds.

The implication is that reserve share and systemic centrality can move in opposite directions for years without contradiction. A reserve share drifting from the high 50s toward the low 50s over a decade is fully compatible with the dollar staying on roughly nine-tenths of FX turnover, because that second number is set by the funding stock and the vehicle-currency equilibrium, not by portfolio preference.

Conditions That Would Turn This Around

This framework is a description of current structure, not a prediction, and there are specific conditions under which it stops describing anything useful. Each deserves the same weight as the base case.

• A euro-area common safe asset at scale. If joint issuance reached even a fifth of the $31.5 trillion Treasury market with comparable secondary liquidity, the reserve-manager constraint would genuinely loosen. The 2026 issuance data shows demand exists. The instrument does not yet.
• A convertible capital account in China. The renminbi's 1.99% reserve share is a policy outcome, not a market verdict. If restrictions were lifted, the trade-finance foothold could convert into reserve demand faster than the gradualist reading implies. There is no sign of this, but the sequence is short if it starts.
• A shock to Treasury market functioning. The dollar's position rests on Treasuries clearing in size under stress. A sustained failure of that assumption — repeated failed auctions, a persistent breakdown in dealer intermediation — would do more damage in a quarter than a decade of reserve diversification.
• Broad sanctioning of reserve assets. Immobilising official claims raises the expected cost of holding them. That effect operates on Layer 1 quickly and on Layer 4 barely at all, but repeated use would compound.
• Payment technology that removes the network problem. If bilateral settlement became cheap enough that thin currency pairs no longer carried a liquidity penalty, the vehicle-currency logic underpinning the 89.2% figure would weaken. Central bank digital currency projects gesture at this. None has yet demonstrated it at cross-border scale.

Note what would not falsify the framework: another two or three points off the COFER share, another year of record gold buying, or further CIPS participant growth. All are consistent with the structure described here.

A Precedent Worth Remembering

The historical case that fits best is not 1971. It is 1978 and 1979, when the US Treasury issued debt denominated in West German marks and Swiss francs — the so-called Carter bonds — to acquire foreign currency for the Exchange Stabilization Fund and defend the dollar. A country whose currency is supposedly unassailable does not borrow in someone else's money.

The instructive part is what followed. The dollar's international role was not durably damaged, the reserve system did not reorganise around the mark, and within a few years the direction had reversed. Acute currency stress and structural currency status proved to be separate variables, and confusing them produced bad forecasts in both directions. The same confusion is available today, in the opposite sign.

What to Watch Next Week

• Swift's monthly currency tracker. The trade finance split is the highest-frequency read on invoicing behaviour. A renminbi share moving durably above 10% would be a genuine change in Layer 2; monthly noise of half a point would not.
• The Fed's H.4.1 release. Central bank liquidity swap balances and FIMA repo usage. Standing dollar swap lines exist with the Bank of Canada, Bank of England, Bank of Japan, ECB and Swiss National Bank. Any non-zero drawdown indicates offshore funding stress in progress, which is a different signal from structural diversification.
• Cross-currency basis for EUR/USD and USD/JPY. The cleanest daily proxy for whether the $4 trillion-a-day swap market is clearing comfortably. Widening beyond typical quarter-end ranges is the tell.
• Treasury auction tails and dealer take-down. Bid-to-cover on its own is a weak signal. The share allotted to primary dealers is the better one, because a rising share means end demand fell short.
• Monthly official gold reporting. Continued accumulation confirms Layer 1 hedging. It informs no judgement about Layers 2 to 4.

Concrete Framework

A monitoring checklist for anyone trying to separate the reserve-composition story from the systemic-role story.

1. Label every claim by layer first. Reserves, invoicing, rails, or funding. A claim about one layer is not evidence about another. Most published de-dollarization analysis fails at this step.
2. Adjust COFER for valuation before reading the level. The IMF flagged that roughly half the 2026 Q1 increase was exchange rate driven. Quarterly moves under about one percentage point should be treated as noise until the valuation decomposition is checked.
3. Track the funding stock quarterly, not the reserve share. The BIS $14.7 trillion series, growing 7.3% year on year, is the tightest available measure of how much of the world is contractually short dollars. It has been rising, not falling.
4. Watch turnover for structural change, not price. The BIS Triennial runs every three years, next in April 2028. Between surveys, the vehicle-currency share is effectively a constant; treat intra-period claims about it with suspicion.
5. Distinguish stress flows from status flows. Dollar strength in a risk-off episode is usually swap-market deleveraging, not an upgrade of US fundamentals. Both produce the same chart; only the swap basis and swap line usage separate them.
6. Set falsification conditions in advance. Write down what would have to happen for the framework to be wrong — a euro common safe asset above roughly $6 trillion, renminbi convertibility, or sustained Treasury market dysfunction — and check against those, not against headlines.
7. Assign probabilities, then revisit on a schedule. The honest position on a ten-year horizon is wide uncertainty on the reserve share and much narrower uncertainty on the funding share. Quarterly review suffices; these variables do not move faster.

The short version: the dollar's reserve share is genuinely contested and may keep drifting lower. Its role as the unit in which the world borrows, hedges and settles is a different variable, on different series and a different clock, currently pointing the other way. Treating them as one story is the most common error in this debate.

This article is macroeconomic analysis for general information. It is not investment, financial or trading advice, and it does not constitute a recommendation to buy or sell any security or currency.