🧭The Probability Desk — Weekend Flagship. A paid-grade, probability-weighted scenario report for long-horizon capital. This week: the slow unpegging of the world's reserve system. Editorial scenario analysis only — not investment, actuarial, or geopolitical advice.
Executive summary
In the fourth quarter of 2025, the U.S. dollar's share of the world's allocated foreign-exchange reserves fell to 56.8% — its lowest reading since the mid-1990s, and the continuation of a quarter-century erosion that has taken the dollar from roughly 71% of reserves in 1999 to under 57% today. The decline is not loud. There is no crisis, no run, no replacement. The dollar remains, by a wide margin, the single most important reserve asset on earth. But the direction is unmistakable, the pace is now being set by the people who hold the reserves rather than by the markets that price them, and the destination is no longer the renminbi. It is gold and a widening basket of "nontraditional" currencies.
For a universal asset owner — a sovereign fund, a public pension, a reserve manager, an insurer with a multi-decade liability — this is not a trading question. It is a question about the denominator of the entire portfolio: the currency in which the risk-free rate is quoted, in which the deepest collateral pool is held, and against which every other claim is measured. The reserve order is the substrate beneath the substrate. When it moves slowly, it moves everything slowly with it.
We frame one resolvable question, build a prior from the actual historical record of reserve-currency transitions, update it against the evidence that has accumulated over the past two years, and run a Monte Carlo simulation of the dollar's reserve share calibrated to twenty-six years of reported data. Our central finding: the most likely path is not a collapse and not a reversal, but a managed drift — a continued grind into the low-50s by the early 2030s, with gold and a diversified currency basket absorbing the outflow. We weight that base case at 55%. We treat an accelerated diversification (25%) as the most consequential alternative, a dollar reconsolidation (12%) as the tail that consensus underweights at its peril, and an outright fragmentation of the system into blocs (8%) as the genuinely disruptive tail.
Desk view, in one paragraph
The dollar is not being dethroned; it is being diversified around. The marginal reserve dollar is no longer being added — it is being rotated into gold (a sanction-proof, issuer-less asset that central banks bought to the tune of roughly 860 tonnes in 2025 — down from 2024's pace but still well above the 2010s norm) and into a long tail of mid-sized currencies. The renminbi is conspicuously not the beneficiary; its reserve share has stalled near 2% and fallen since 2022. The Desk reads this as the single most important and least dramatic macro fact of the decade: a world that is de-risking from the dollar without yet agreeing on what to de-risk into. That ambiguity — diversification without a successor — is itself the regime, and it favours real assets, gold, duration discipline, and currency breadth over any single replacement bet.
The situation as of today
As of late May 2026, the picture across the reserve system reads as follows. The International Monetary Fund's reserve-composition data put the dollar at 56.8% of allocated reserves at end-2025, down from 56.9% the prior quarter and from 57.8% at the start of 2025. The euro sits near 21%; the renminbi near 2.0% and drifting lower; the residual "other" bucket — Canadian and Australian dollars, the Korean won, the Singapore dollar, the Nordic and Swiss currencies — has climbed past 20% and is now larger than every non-dollar, non-euro currency combined was a decade ago. Total world reserves stand at roughly $13.1 trillion.
Gold has done the rest of the work. Official-sector gold buying ran at about 863 tonnes in 2025 — down roughly 21% from 2024 and the lowest since 2021, yet still close to double the 2010–2021 average near 470 tonnes, and a fourth consecutive year above 800 tonnes — and continued into 2026, with several central banks (Poland prominent among them) leading the additions. The metal itself trades near $4,500 an ounce after touching an all-time high above $5,500 in late January 2026, and is up roughly a third year-on-year. The U.S. Federal Reserve's own broad dollar index sits near 119, off its highs, while the 30-year Treasury yield has pushed above 5% — a fiscal backdrop that frames, rather than triggers, the reserve question.
What changed
Three things crystallised in the past two years and continued to register in the most recent data. First, the IMF's late-2025 reserve update confirmed the dollar's break below 57% on a like-for-like basis after a methodology revision that, helpfully, eliminated the old "unallocated" bucket and now accounts for 100% of reported reserves — removing a long-standing excuse to wave away the trend. Second, the World Gold Council's 2025 survey of central banks recorded the most diversification-minded readings in its eight-year history: 95% of respondents expected global official gold reserves to rise over the following year, a record 43% planned to add to their own holdings (up from 29% in 2024), and 76% expected gold to hold a higher share of reserves in five years. Third, the OMFIF Global Public Investor survey found the dollar was the only reserve currency to see net demand fall, while a net 16% of reserve managers planned to add euros (up from 7%) and close to 60% described themselves as actively seeking to diversify within two years. The signal is consistent across three independent institutional surveys: the holders of reserves are telling us, in their own words, what they intend to do.
Source ledger
Thirty-one sources underpin this report. Confidence is graded H/M/L. "Moves model" flags whether the source materially shifted a probability.
| # | Source | Date | Data point used | Conf. | Moves model |
|---|---|---|---|---|---|
| 1 | IMF COFER data brief (currency composition of reserves) | Mar 2026 (Q4 2025 data) | USD share 56.77%; total reserves $13.14T | H | Yes |
| 2 | IMF blog, "Dollar's share held steady when adjusted for FX moves" | Oct 2025 | Valuation vs flow decomposition of share decline | H | Yes |
| 3 | IMF Working Paper 2022/058, "The Stealth Erosion of Dollar Dominance" (Arslanalp, Eichengreen, Simpson-Bell) | 2022 | Erosion to nontraditional currencies; ~71%→59% over two decades | H | Yes |
| 4 | Federal Reserve, "The International Role of the U.S. Dollar — 2025 Edition" | Jul 2025 | Dollar invoicing, debt, FX-turnover dominance intact | H | Yes |
| 5 | Federal Reserve IFDP 1420, central-bank gold purchases & dollar role | 2025 | Gold accumulation as diversification, not pure de-dollarization | H | Yes |
| 6 | NY Fed Staff Report 1087, "Drivers of Dollar Share in FX Reserves" | 2024 | Inertia / network effects in reserve composition | H | Yes |
| 7 | NBER WP 34478, "Our Underappreciated International Reserve System" | 2026 | Structure and resilience of the multi-currency system | M | Yes |
| 8 | World Gold Council, Central Bank Gold Reserves Survey 2025 | Jun 2025 | 95% expect global gold reserves up; 43% plan own increases; 76% see higher gold share in 5y | H | Yes |
| 9 | World Gold Council, Gold Demand Trends — Full Year 2025 | Feb 2026 | Official sector ~863t purchased in 2025 (down ~21% y/y; still >800t) | H | Yes |
| 10 | World Gold Council, central-bank gold statistics (K. Gopaul) | 2026 | 2026 additions led by Poland; continued >monthly buying | M | Yes |
| 11 | OMFIF Global Public Investor 2025 | Jun 2025 | Dollar only currency with falling demand; net +16% euro; ~60% diversifying | H | Yes |
| 12 | OMFIF, "Central banks turn to gold over the dollar" | Jun 2025 | Gold as preferred diversifier; neutrality rationale | M | No |
| 13 | OMFIF, "Gold's status from theoretical to tactical hedge" | Jun 2025 | Reserve-manager reclassification of gold | M | No |
| 14 | U.S. Federal Reserve H.15 / Treasury data (30y, 10y yields) | 27 May 2026 | 30y 5.01%, 10y 4.48% | H | No |
| 15 | U.S. Federal Reserve broad dollar index | 22 May 2026 | Broad USD index ~119.3 | H | No |
| 16 | Fortune / market data, gold spot | 26–28 May 2026 | Gold ~$4,500/oz; +~36% y/y; Jan-2026 peak ~$5,590 | M | No |
| 17 | IMF COFER FAQ / methodology note | 2025 | Unallocated bucket eliminated from Q3 2025, revised to 2000 | H | Yes |
| 18 | Bank for International Settlements Triennial FX survey (latest) | 2022/2025 | Dollar on ~88% of FX trades — turnover dominance | H | No |
| 19 | SWIFT RMB Tracker | 2025–26 | RMB payments share low-single-digits; dollar/euro dominant | M | Yes |
| 20 | Reuters / Bloomberg reserve and gold coverage | 2025–26 | Corroborating flows and central-bank commentary | M | No |
| 21 | People's Bank of China gold reserve disclosures | 2025–26 | Resumed reported buying after pauses | M | Yes |
| 22 | National Bank of Poland communications | 2025–26 | Target gold share ~20% of reserves | M | Yes |
| 23 | European Central Bank, international role of the euro | 2025 | Euro share stable ~20–21%; no euro "moment" | H | Yes |
| 24 | U.S. Treasury, frozen Russian reserve assets context | 2022–26 | ~$300bn immobilised — the weaponization catalyst | H | Yes |
| 25 | IMF World Economic Outlook (fiscal/debt) | Apr 2026 | U.S. debt trajectory; global fiscal backdrop | H | No |
| 26 | World Bank / IDS external-debt data | 2025 | EM dollar-debt stock — switching-cost evidence | M | No |
| 27 | Invesco Global Sovereign Asset Management Study | 2025 | SWF/reserve diversification intentions; gold appetite | M | Yes |
| 28 | Atlantic Council Dollar Dominance Monitor | 2025–26 | Cross-checks reserve, payment, and sanctions trends | M | No |
| 29 | WGC, "Gold's bull run" / ING commentary | 2026 | 2026 price and demand outlook | L | No |
| 30 | BestBrokers reserve-landscape compilation of COFER | 2025–26 | Quarter-by-quarter share series (secondary) | L | No |
| 31 | Federal Reserve / academic estimates of reserve-share volatility | 2024–25 | Calibration of drift and variance for the simulation | M | Yes |
Key data table
| Variable | Current | Prior | Dir. | Source | Conf. |
|---|---|---|---|---|---|
| USD share, allocated reserves | 56.8% (Q4 2025) | 56.9% (Q3 2025) | ↓ | IMF COFER | H |
| USD share, start of 2025 | — | 57.8% (Q1 2025) | ↓ | IMF COFER | H |
| USD share, 1999 reference | — | ~71% | ↓ (26y) | IMF / IMF WP 22/058 | H |
| Euro share | ~21% | ~20% | →/↑ | IMF COFER / ECB | H |
| Renminbi share | ~2.0% | ~2.8% (2022 peak) | ↓ | IMF COFER | H |
| "Other" (nontraditional) share | ~20%+ | ~20.1% | ↑ | IMF COFER | H |
| Total world FX reserves | ~$13.1T | ~$13.0T | ↑ | IMF COFER | H |
| Official gold buying, annual | ~863t (2025) | ~1,045t (2024) | ↓ (still >800t) | WGC | H |
| Gold spot | ~$4,500/oz (May 26) | ~$5,590 peak (Jan 26) | ↓ from peak, ↑ y/y | Market data | M |
| Central banks planning to add gold (own) | 43% (2025) | 29% (2024) | ↑ | WGC survey | H |
| Expect higher gold share in 5y | 76% | 69% | ↑ | WGC survey | H |
| U.S. 30-year Treasury yield | 5.01% (27 May) | 5.07% (22 May) | → | Fed/Treasury | H |
| Broad USD index | ~119.3 (22 May) | ~119.2 | → | Federal Reserve | H |
The forecast question
Primary, resolvable (12-month): Will the U.S. dollar's share of allocated global FX reserves, as reported by the IMF, print below 55.0% in any quarter through Q1 2027, and will official-sector net gold purchases exceed 800 tonnes in calendar 2026?
Resolution criteria: the first leg resolves on IMF COFER quarterly releases through the Q1-2027 print; the second on the World Gold Council's official-sector demand tally for full-year 2026. Both legs must clear for a "yes."
Structural framing (25-year): conditioned on no global financial crisis, where does the dollar's reserve share sit in 2035 and 2050, and what is the probability it falls below 50% (loses outright majority) within each horizon?
Base rate: the modern dollar share has declined at an average of roughly 0.5 percentage points a year since 1999, with quarterly noise of similar magnitude driven by exchange-rate valuation. A sub-55% print within twelve months is therefore consistent with trend but not assured; official gold buying has exceeded 800 tonnes in each of the last four years (about 863 tonnes in 2025, down from 2024 but comfortably above the threshold).
What would change the call: a global risk-off episode that drives a flight into dollar liquidity (raises the share via valuation and flow); a credible euro-area fiscal-union step (gives the euro a genuine alternative); or a U.S. fiscal/political shock that accelerates active diversification.
Establishing the prior — analogues and base rates
Reserve-currency transitions are rare, slow, and asymmetric. The relevant analogues:
Sterling to dollar (1914–1956). The dollar overtook sterling in economic weight by the 1910s but did not clearly displace it in reserves until after the Second World War — a four-decade lag between economic and monetary leadership. Why relevant: incumbency is extraordinarily sticky; the new leader must offer depth, not just size. Why possibly misleading: that transition had a single clear successor; today's does not.
The 1970s dollar crisis and the end of Bretton Woods. The dollar's convertibility ended in 1971; reserves diversified, gold soared, and yet the dollar's reserve role survived and recovered. Implication for the prior: a falling share is not the same as a failing currency, and reserve roles can stabilise after shocks.
The euro's launch (1999–2010). The euro rose from nothing to a peak near 28% of reserves by 2009, then stalled and partly reversed through the sovereign-debt crisis. Implication: a credible alternative can take share quickly — and lose it just as quickly when its own fault lines show.
The 2022 weaponization of reserves. The immobilisation of roughly $300bn of Russian reserve assets was, for every non-aligned reserve manager, a live demonstration that dollar (and euro) reserves carry political risk. Implication: this is the regime-change event that distinguishes today's prior from the gentle 2000s drift — it added a one-directional, price-insensitive bid for issuer-less gold.
Synthesising: the base-rate frequency of a reserve currency losing outright majority within a decade, absent a war or a clearly superior successor, is low. But the inside-view evidence — surveys, gold flows, the sanctions precedent — argues the pace has structurally quickened from its 2000s norm. We hold the prior at "continued slow erosion" and let the evidence update it.
Updating the model — the evidence table
Editorial standard: we publish no probability without showing the evidence that moved it. The base case begins at the historical-trend prior (a continued ~0.5pp/yr drift, implying the share stays in the mid-50s over twelve months and grinds toward the low-50s by the early 2030s) and is updated as follows.
| Evidence | Direction | Strength | Effect on view | Conf. |
|---|---|---|---|---|
| WGC 2025 survey: 43% of central banks plan to add gold (vs 29%); 76% see higher gold share in 5y | Faster diversification | Strong | Raises weight on accelerated case; supports >800t gold leg | H |
| OMFIF GPI 2025: dollar the only currency with falling net demand; ~60% diversifying | Faster diversification | Strong | Confirms holder intent, not just price effects | H |
| IMF: ~half the recent share decline is FX valuation, not active selling | Slower / stabilising | Moderate | Caps the base-case drift; argues against collapse narrative | H |
| Renminbi share stalled ~2% and falling since 2022 | No successor | Strong | Rules out a fast RMB-led displacement; favours "diversification without a successor" | H |
| Network effects / inertia (NY Fed): dollar debt, invoicing, collateral depth intact | Slower | Strong | Anchors a high reconsolidation floor; lowers fragmentation odds | H |
| 2022 reserve immobilisation precedent | Faster diversification | Moderate-strong | One-directional, price-insensitive gold bid for non-aligned holders | H |
| U.S. fiscal trajectory; 30y yield >5% | Faster (long-run) | Moderate | Raises term premium and questions over dollar-asset supply; slow burn | M |
| Euro area still without fiscal union; no euro "moment" | Slower for alternatives | Moderate | Keeps the outflow going to gold + small currencies, not a rival bloc | H |
Posterior: the weight of evidence tilts the pace up from the gentle 2000s drift (surveys + gold flows + sanctions precedent) but is held back from any collapse narrative by valuation mechanics, the absence of a successor, and deep network effects. Net, we keep a base case of continued managed drift but raise the probability of acceleration relative to a naïve trend extrapolation.
Scenarios
Four mutually exclusive, collectively exhaustive regimes for the reserve order over the next decade. Weights sum to 100%, rounded to 5%.
| Scenario | Prob. | Narrative | Trigger / tripwire | What falsifies it |
|---|---|---|---|---|
| 1. Managed drift (base) | 55% | Stealth erosion continues near historical pace; dollar share grinds to the low-50s by the early 2030s but stays the single largest reserve asset; gold and nontraditional currencies absorb the flow; no abrupt break. | COFER prints continue 0.3–0.7pp/yr lower; gold buying stays 700–1,100t/yr. | Share stabilises >57% for a year, or falls below 50% before 2030. |
| 2. Accelerated diversification | 25% | A catalyst — further reserve weaponization, a U.S. fiscal/political shock, or secondary-sanctions overreach — speeds the shift. Dollar share breaks below 50% before 2030; gold share of reserves pushes toward 25–30%; official buying stays >1,000t/yr. | Two consecutive COFER prints >0.8pp lower; gold buying >1,300t in a year; new immobilisation event. | Drift stays at trend; no fresh weaponization; gold buying cools below 700t. |
| 3. Dollar reconsolidation (tail) | 12% | A global risk-off or crisis reasserts dollar-liquidity primacy; share stabilises or ticks up on valuation and flight-to-quality; gold buying cools as the opportunity cost of reserves rises. | Share rises >58% in a stress quarter; net official gold sales appear. | Share keeps falling through a risk-off episode (would signal a true regime break). |
| 4. Monetary bloc-ification (tail) | 8% | The system bifurcates into a dollar bloc and a non-aligned bloc settling more in gold, bilateral lines, and local currencies; reported reserve data becomes a less complete picture as off-book settlement rises. | Material rise in non-dollar bilateral settlement + CIPS/alt-rail volumes; gold repatriation wave. | Non-aligned trade keeps clearing in dollars; no durable alternative rail scales. |
Simulation — results
We ran a Monte Carlo simulation of the dollar's allocated reserve share: 50,000 paths, quarterly steps, over a 25-year horizon from the Q4-2025 starting value of 56.8%. The model is a drift-diffusion with jumps. Per-path annual drift is drawn from a normal distribution centred on −0.50pp/yr (the historical pace) with a 0.30pp standard deviation to represent uncertainty about the rate of change. Quarterly diffusion volatility is 0.55pp, calibrated to the observed quarter-to-quarter variance of the reported share (which is dominated by FX valuation). Two jump processes are layered on: a "diversification shock" (roughly one in seven years; mean −1.6pp; sanctions/geopolitical breaks) and a rarer "reconsolidation shock" (flight to dollar liquidity; mean +1.2pp). Shares are bounded to a structural floor reflecting network effects.
| Output | Result |
|---|---|
| P(share prints <55.0% within next 4 quarters) | 21% |
| P(share prints <56.0% within next 4 quarters) | 54% |
| 2030 share — P10 / P50 / P90 | 49.3% / 53.6% / 57.8% |
| 2035 share — P10 / P50 / P90 | 43.8% / 50.4% / 56.9% |
| 2050 share — P10 / P50 / P90 | 28.0% / 40.9% / 53.5% |
| P(loses outright majority, <50%, by 2035) | 53% |
| P(<50% by 2050) | 87% |
| P(<40% by 2050) | 50% |
| P(reconsolidates >58% and ends 2050 above current level) | 6% |
| Median year share first crosses below 50% | ~2034 |
Reading the simulation. The central path is not a cliff. By 2030 the model's median dollar share is in the low-50s; the loss of outright majority is roughly a coin-flip by 2035 and a near-certainty by 2050 — but "loss of majority" is not "loss of primacy." Even at the 2050 median of ~41%, the dollar remains comfortably the largest single reserve asset in a world where no rival exceeds the low-20s. The simulation's tails are instructive: a 6% mass reconsolidates above today's level (the crisis-flight scenario), and a 10th-percentile path reaches the high-20s by 2050 (the accelerated/fragmentation tail). The near-term resolvable question's first leg — a sub-55% print within a year — sits at 21% in the model, i.e. plausible but below even odds; the gold leg is, on the survey evidence, far more likely than not.
Honesty note. This is a model of the reported reserve share, calibrated to its own history. It deliberately does not attempt to forecast politics, wars, or a euro fiscal union; those enter only through the jump intensities, which are themselves uncertain. The percentiles are model outputs, not predictions, and the value of the exercise is the shape of the distribution — slow drift, fat diversification tail, thin reconsolidation tail — not any single number.
Market pricing vs. the Desk view
Gold near $4,500, off a $5,590 January peak, prices a great deal of the diversification story already; the metal is no longer cheap insurance. The dollar, by contrast, still trades and funds as though its reserve primacy were permanent — dollar credit spreads, the depth of the Treasury collateral market, and the 88%-of-FX-trades turnover all reflect incumbency, not erosion. The Desk's highest-conviction mispricing is therefore not gold and not the dollar spot rate, but long-dated dollar duration relative to the term premium a slowly diversifying buyer base should demand. If the marginal foreign official buyer is rotating into gold and small currencies rather than adding Treasuries, the structural bid that suppressed the U.S. term premium for two decades is fading — and a 30-year yield above 5% is the early tell, not the end state. What consensus is missing: it keeps asking "what replaces the dollar?" (answer: nothing soon) and so concludes "nothing is happening." The right question is "who funds the incremental Treasury at the old price?" — and the answer is increasingly "a domestic or price-sensitive buyer, at a higher yield." What would prove the Desk wrong: a durable compression of the term premium alongside a falling reserve share, which would show the diversification is happening without any cost to dollar-asset funding.
Universal-owner portfolio map
Strategic implications for a diversified, long-horizon owner — not personalised advice. Cells read as the directional pressure on each exposure under each regime over the stated horizon.
| Exposure | Base (drift) 12m | Base 5y | Accelerated 5y | Reconsolidation 5y |
|---|---|---|---|---|
| Gold / monetary metals | Supportive | Supportive | Strongly supportive | Headwind |
| Long-duration U.S. Treasuries | Neutral–soft | Soft (term premium ↑) | Negative | Strongly positive |
| U.S. dollar cash / FX | Neutral | Soft | Negative | Positive |
| Euro & euro assets | Neutral | Mild support | Support | Soft |
| Nontraditional FX (AUD, CAD, KRW, SGD) | Mild support | Support | Strong support | Soft |
| EM local-currency debt | Neutral | Mild support | Support | Negative |
| Real assets / infrastructure / commodities | Support | Support | Strong support | Mixed |
| Global equities (USD-denominated) | Neutral | Mild FX drag | FX drag, earnings mixed | Support |
| Inflation-linked bonds | Support | Support | Strong support | Neutral |
The through-line for a universal owner: a slowly diversifying reserve order argues for currency breadth (not a single replacement bet), a structurally larger strategic allocation to gold and real assets than the 2010s playbook implied, discipline on unhedged long-dollar duration, and explicit recognition that the home-currency "risk-free" anchor is itself drifting. None of these require a view that the dollar fails — only that it is, gradually, shared.
Second- and third-order effects
U.S. funding costs. A fading official bid raises the term premium and shifts incremental Treasury funding to domestic and price-sensitive buyers — a slow tightening of the fiscal constraint that compounds over a decade. Gold's monetary re-rating. If gold's share of reserves moves from the low-20s toward 30%, the metal is being repriced as a monetary asset, not a commodity — with spillovers to mining capex, central-bank vaulting and repatriation, and the gold-lease market. Sanctions architecture. Every use of reserve immobilisation raises the marginal non-aligned holder's diversification rate; the tool's deterrent value and its dedollarizing side-effect are now in tension. Euro-area incentives. A persistent diversification flow is the strongest external argument for joint euro safe-asset issuance; a credible step there would redirect the outflow from gold to euros and reshape the whole map. Emerging-market resilience. Reserve diversification by EM central banks (gold + local currencies) modestly reduces their dollar-funding fragility — but the vast stock of EM dollar debt keeps the switching cost high. Global South settlement. Bilateral and local-currency trade settlement grows at the margin without yet threatening dollar invoicing — the gap between settlement experiments and reserve reality is the single most over-interpreted data series in this debate.
What we're watching
| Indicator | Why it matters | Threshold that moves the model |
|---|---|---|
| IMF COFER quarterly USD share | The headline series | <55% print → confirms accelerated leg; >58% → reconsolidation |
| WGC official-sector gold demand | The active diversification proxy | >1,300t/yr → accelerated; <700t/yr → cooling |
| Gold price & lease rates | Pricing of the monetary re-rating | New highs on official buying vs investor selling |
| U.S. 30-year term premium | The funding-cost transmission | Sustained rise alongside falling reserve share |
| Renminbi reserve share | Tests the "no successor" thesis | Sustained rise above ~3% → successor risk |
| Euro reserve share / EU joint issuance | The only credible rival path | Joint safe-asset step → redirects the flow |
| New reserve-immobilisation events | The diversification catalyst | Any fresh major freeze → raises accelerated weight |
| Foreign official Treasury holdings (TIC) | Direct read on the official bid | Sustained decline in official holdings share |
| Non-dollar bilateral settlement / CIPS volume | The fragmentation tripwire | Step-change in non-dollar clearing |
| SWF strategic gold/real-asset allocations | The audience's own behaviour | Named large funds adding gold mandates |
Red-team — how this could be wrong
The decline is mostly valuation, not diversification. The IMF itself notes that, adjusted for exchange-rate moves, the dollar's share has been far steadier than the headline implies. If a stronger dollar reverses the valuation effect, the reported share could stabilise or rise even as nothing structural changes — and our base case's drift would prove too fast.
Network effects are stronger than we modelled. The dollar's role in invoicing, debt issuance, FX turnover and collateral is self-reinforcing; the NY Fed's work suggests reserve shares are extremely persistent. Our floor may be too low and our drift too steep.
"No successor" cuts the other way. A reserve system with no alternative may simply stay dollar-centric out of necessity, parking diversification in gold without ever materially lowering the dollar's working role. In that case the COFER share keeps drifting but the dollar's functional dominance is untouched — making the whole question less consequential than the headline suggests.
The gold trade is crowded. At ~$4,500 after a parabolic run, gold prices much of the diversification story. If official buying cools and investor positioning unwinds, the most visible symptom of the thesis could correct sharply — without invalidating the slow structural drift beneath it.
Weakest assumption: that the historical −0.5pp/yr drift is a reasonable centre for the next decade. It is an average across a benign period and a weaponization period; the true process is almost certainly regime-dependent, and our jump terms are a crude proxy for that.
Methodology box
This report follows the Probability Desk's flagship standard: a stated, resolvable question; a prior built from historical analogues and base rates; a Bayesian-style evidence update that shows what moved the view; a mutually-exclusive, collectively-exhaustive scenario set weighted to 100%; and a real Monte Carlo simulation (50,000 paths) of the reserve share, calibrated to twenty-six years of reported data, with all assumptions disclosed. Figures are drawn from primary institutional sources — the IMF, the World Gold Council, OMFIF, the Federal Reserve, the ECB and the BIS — and cited with dates and confidence grades; no probability is published without the evidence behind it. Where a series could not be independently confirmed it is graded low confidence. The simulation forecasts the reported share, not geopolitics; its outputs are distributions, not predictions. The Desk owns the numbers; the model is the engine, not the author.
Reviewed under the Probability Desk's independent pre-publication scenario-review process.
Disclaimer
This report is for informational and research purposes only and does not constitute investment, legal, tax, or fin