NY Fed Says Dollar's Reserve Retreat Is a Few Managers' Doing

Research from the Federal Reserve Bank of New York, reported on September 7, 2026, concludes that the decline in the dollar's share of global foreign-exchange reserves reflects the actions of a small number of reserve managers rather than a broad reallocation away from the currency.

The long-running story that the world's central banks are quietly walking away from the US dollar has just been handed a significant qualification. Research from the Federal Reserve Bank of New York concludes that the fall in the dollar's share of global foreign-exchange reserves does not reflect a broad, coordinated reallocation by reserve managers around the world. Instead, according to the study reported by Bloomberg Economics, the decline is attributable to moves by a small number of reserve managers.

That distinction matters more than it may sound. A headline aggregate number — the dollar's percentage share of allocated global reserves — behaves very differently depending on whether it is falling because almost everyone is trimming a little, or because a handful of large holders are trimming a lot. The first is a structural verdict on the currency. The second is a set of country-specific decisions that happen to move a global average.

Why the aggregate share is a noisy signal

Reserve composition data is one of the most heavily cited series in international finance, and one of the most frequently misread. It is a share, not a stock, which means it can move for at least three reasons that have nothing to do with a central bank selling dollars.

• Valuation effects. When the dollar weakens against the euro, yen or sterling, the non-dollar portion of a portfolio is worth more in dollar terms. The dollar's share falls even if not a single bond has been sold.
• Composition of the reporting universe. Not every country discloses the currency breakdown of its reserves. Changes in who reports, and how much they hold, shift the aggregate.
• Concentration. Reserve holdings are extraordinarily lopsided. A small group of very large holders can pull the global average in their direction on their own.

The New York Fed's finding speaks to that third channel. If the decline in the dollar share is being driven by a small number of managers, then the average is describing those managers rather than describing a consensus. Extrapolating from it to "the world is de-dollarising" is a category error.

What a concentrated shift changes about the risk

For anyone holding US assets, the difference between broad and concentrated reallocation is the difference between a slow structural bid disappearing and a few identifiable sellers doing something for their own reasons — sanctions exposure, trade invoicing patterns, domestic politics, or simply a decision to hold more gold.

A broad-based exit would imply persistent, mechanical selling pressure on Treasuries and a steady erosion of the dollar's role in invoicing, settlement and private-sector borrowing. That is the version that shows up in the more dramatic commentary. A concentrated shift implies something more limited: the aggregate line on the chart is real, but the behaviour behind it is not general, and it is not obviously self-reinforcing.

It also changes what to watch. If the move is concentrated, the relevant question is not "what is the global reserve manager doing" but "what are these specific reserve managers doing next, and how much more do they have left to reallocate." A handful of large holders can only rotate out of dollars once. Their capacity to keep pushing the aggregate lower is finite in a way that a genuine global reallocation would not be.

The reserve currency argument keeps outrunning the evidence

De-dollarisation has become one of the most durable market narratives of the decade, invoked to explain gold's strength, emerging-market central bank buying, the rise of bilateral settlement arrangements and the ambitions attached to various currency blocs. Some of that is grounded. But the narrative has repeatedly leaned on a single aggregate statistic doing work it was never designed to do.

The New York Fed's contribution is essentially forensic: decompose the aggregate, and the story looks smaller. It does not say the dollar's share has not fallen. It says the fall does not evidence what it is routinely used to evidence.

There is a practical reason central banks are sticky here. Reserves exist to be deployed in a crisis, which means they need to be held in the deepest, most liquid, most reliably convertible market available. For most reserve managers that remains the US Treasury market, whatever their views on US fiscal policy or the use of financial sanctions. Diversification at the margin is cheap; wholesale relocation of a war chest is not.

Where US markets stood going into the study

The research lands against a US equity tape that has been trading off rates and jobs data rather than currency composition. In the most recent session before publication, ending Friday, September 4, 2026, the S&P 500 tracker (NYSEARCA: SPY) closed at $770.19, down 0.39% from the prior close of $773.17, with a day range of $769.00 to $772.87. The Dow 30 fund (NYSEARCA: DIA) finished at $534.08, off 0.53% from $536.93. The Nasdaq 100 tracker (NASDAQ: QQQ) was the exception, closing at $718.96, up 0.18% from $717.67 after ranging between $716.56 and $721.86.

That split — large-cap tech firm, broad market and industrials softer — is the signature of a rate-sensitive session, not a currency-regime session. It is a useful reminder that reserve-composition trends operate on a timescale measured in years, and rarely explain a given day's close.

What to watch from here

Three things will determine whether the New York Fed's framing holds up.

• Whether the concentrated sellers continue. If the same small group keeps reallocating, the aggregate keeps falling and the debate about causes becomes academic for anyone marking to market.
• Whether other managers follow. The study's whole point is that they have not, in a broad way. Evidence that mid-sized reserve holders are joining in would flip the conclusion.
• Valuation versus flow. Future reserve data should be read with the dollar's own path in mind. A weaker dollar mechanically shrinks its reported share; a stronger one flatters it. Separating price from flow is the entire analytical exercise.

For investors, the useful takeaway is not that de-dollarisation is a myth. It is that the single number most often used to prove it is a weighted average dominated by a few participants, and that a weighted average is not a referendum. Anyone building a portfolio thesis on the dollar's reserve status should now be asking who, specifically, is selling — and whether they have much left to sell.

Key facts

• Source of research: Federal Reserve Bank of New York
• Core finding: Dollar's falling FX reserve share reflects a small number of reserve managers, not broad reallocation
• S&P 500 tracker (SPY): $770.19, -0.39%, close of Fri, 04 Sep 2026 20:00 GMT
• Nasdaq 100 tracker (QQQ): $718.96, +0.18%, close of Fri, 04 Sep 2026 20:00 GMT

Frequently asked questions

What did the New York Fed study actually conclude?

The Federal Reserve Bank of New York found that the decline in the dollar's share of global foreign-exchange reserves does not reflect a widespread reallocation away from the currency by reserve managers generally. The research attributes the decline instead to moves by a small number of reserve managers, meaning a concentrated set of decisions is pulling down a global aggregate figure.

Why does it matter whether the shift is broad or concentrated?

A broad reallocation would suggest a structural loss of confidence in the dollar with persistent selling pressure behind it. A concentrated shift means a few identifiable holders acted for their own reasons and that their capacity to keep moving the aggregate is finite. The two scenarios carry very different implications for Treasury demand and dollar risk.

Can the dollar's reserve share fall without any central bank selling dollars?

Yes. The figure is a share of a portfolio measured in dollar terms. If the dollar weakens against other reserve currencies, the non-dollar holdings are worth more in dollar terms and the dollar's share falls mechanically. Changes in which countries report their currency breakdown can also shift the aggregate without any transactions occurring.

Does this mean de-dollarisation is not happening?

The research does not claim the dollar's reserve share has not declined. It challenges the interpretation that the decline proves a general, worldwide move out of the currency. The distinction is about what the aggregate statistic can and cannot support as evidence, not about whether the number itself has moved.

Why do central banks tend to keep holding dollars?

Reserves are meant to be deployable in a crisis, which requires the deepest and most liquid market available for rapid conversion. For most reserve managers that remains the US Treasury market. Trimming exposure at the margin is straightforward, but relocating an entire reserve stock into shallower markets carries real operational and liquidity costs.

How did US markets close before this research was reported?

In the last session before publication, ending Friday, September 4, 2026, the S&P 500 tracker SPY closed at $770.19, down 0.39%. The Dow 30 fund DIA finished at $534.08, down 0.53%. The Nasdaq 100 tracker QQQ closed higher at $718.96, up 0.18%, with a day range of $716.56 to $721.86.

Sources

• Global Reserves Shift From Dollar Overstated, NY Fed Study Says — Bloomberg Economics

Photo: Engin Akyurt · Pexels Licence — source