The dollar can lose reserve share without a single sale

IMF headquarters at 700 19th Street NW, Washington, D.C., May 31, 2024. The Fund compiles the reserve-currency statistics examined here. Photograph: Tony Webster / Wikimedia Commons, CC BY 2.0; resized. [7]

A rise in the dollar’s reserve share looks like fresh buying, but the latest IMF release offers a narrower signal: exchange-rate effects explain around half of its first-quarter 2026 increase.[1] My reading is that a reserve-share headline alone gives too little evidence to reprice official demand for dollars or Treasuries.

Data checked September 23, 2026. The latest COFER brief available on the IMF’s dataset page covers the first quarter of 2026; historical figures below are identified by publication vintage.[2]

A portfolio can stand still while its shares move

The IMF’s Currency Composition of Official Foreign Exchange Reserves, or COFER, puts reserve holdings into a common unit: US dollars. Reporting authorities translate assets denominated in other currencies using quarter-end market exchange rates. The dollar share then divides dollar-denominated holdings by the dollar value of the entire foreign-exchange reserve pool.[6]

Imagine a reserve manager whose portfolio contains dollar deposits and euro bonds. Hold the quantities and local-currency prices fixed. If the euro appreciates against the dollar, the euro bonds become worth more when expressed in dollars. The portfolio’s total dollar value rises; the dollar deposits have not changed. Their share consequently falls.

The manager need not have placed an order. The exchange rate has changed the measuring stick.

The reverse works too. A stronger dollar reduces the translated value of foreign-currency assets, lifting the dollar share even with unchanged holdings. This makes the statistic unusually easy to overread: a market move can alter the reserve share subsequently invoked to explain that same market move. That circular inference adds no evidence of buying or selling.

The same mechanism, in opposite directions

In its July 1, 2026 brief, the IMF reported that the dollar’s share rose from 56.42% in the final quarter of 2025 to 57.13% in the first quarter of 2026. It attributed around half of the increase to exchange-rate valuation. Those percentages establish a change in portfolio composition measured at market values; they do not establish the size of a purchase program.[1]

An earlier episode makes the distinction sharper. The IMF’s October 2025 analysis of the second quarter of that year showed a 1.47-percentage-point headline fall in the dollar share. Holding exchange rates constant reduced that decline to just 0.12 percentage points, calculated from the shares in that publication. Currency movements explained 92% of the reported decrease.[3]

These historical figures belong to the series as published at that time. They illustrate the mechanism; they should not be spliced into today’s revised database to manufacture a continuous trend.

Nor does the small exchange-rate-adjusted residual automatically measure net selling. Bond prices move as yields change, altering the value of reserve securities without a transaction. The latest IMF brief explicitly separates this source of valuation change from active management. Removing currency translation is the beginning of a flow analysis, not its completion.[1]

For a Treasury-demand argument, an additional distinction matters: COFER groups assets by currency. Dollar deposits and dollar securities both enter the dollar bucket. A reserve manager could change the balance between them without changing currency exposure. The aggregate dollar share cannot identify which Treasury maturities gained or lost a buyer.[2]

The denominator was rebuilt

Starting with the third-quarter 2025 data, the IMF eliminated COFER’s old “unallocated” category and revised the historical series back to the first quarter of 2000. Previously, currency shares described the allocated portion of reserves; the revised series estimates a currency composition for the full foreign-exchange reserve pool.[4]

The November 2025 technical note explains how staff fill gaps using methods including estimates based on comparable reporting groups and the carrying forward of previous reports. The broad historical trends survive, but individual currency shares shift. The estimated portion remains an estimate, even when the resulting table looks complete.[4]

For readers comparing releases, this creates a practical obligation: use a consistent data vintage. A percentage copied from an old article and one downloaded today may have different statistical foundations. A revised history is evidence of improved measurement, not evidence that reserve managers retroactively changed their portfolios.

There is another boundary around the denominator: monetary gold is outside COFER. A higher gold price can enlarge gold’s share of broader official reserves without directly lowering the dollar share reported in this particular dataset. Stories about gold overtaking another reserve asset and stories about COFER are answering different questions.[6]

Diversification still deserves a hearing

The strongest counterweight is that deliberate diversification is real. In a 2022 study summary, IMF researchers documented a longer-term shift toward nontraditional reserve currencies, including those of smaller economies. They connected the shift to portfolio considerations and improvements that made these currencies easier to trade. Currency translation cannot dismiss a documented allocation strategy.[5]

A central bank may also diversify gradually by directing new reserve accumulation elsewhere. It need not dump its existing dollar assets to reduce their share. That possibility matters for demand at the margin, even when there is no dramatic sale to photograph.

My falsifier for a valuation-led interpretation of a future quarterly move is a persistent change that survives adjustment for exchange rates and security prices, accompanied by disclosed allocation decisions in the same direction. Under those conditions, continued appeals to translation would understate the evidence. COFER’s confidential country submissions mean public aggregates alone cannot identify which authority made the decision.[6]

What would settle the next headline

At the next quarterly COFER release: compare the reported share movement with the IMF’s valuation discussion and examine both quarters in the same updated dataset. A large headline move with little movement after adjustment would strengthen the valuation explanation.[2]
When the accompanying revision and methodology notes appear: check the imputed share and any changes to past observations before attributing a discontinuity to reserve-manager behavior.[2][4]
At the next annual reserve-management disclosures from central banks: look for changed target allocations and explanations of portfolio transactions. Those disclosures can turn an aggregate statistical hint into evidence of a deliberate decision, as the earlier IMF research illustrates.[5]

cronfeed.work