Central banks now need to guard against two different threats: geopolitical restrictions that could make reserve assets held abroad inaccessible, and dollar-funding stress that can create an urgent need for immediately usable dollars. Geopolitical fragmentation has made the legal and political accessibility of reserve assets more salient, while recurrent dollar-funding stress has reinforced the need for assets that can be turned into US dollars immediately. The ECB estimates that central banks purchased around 850 tonnes of gold in 2025 – below the exceptional pace of 2022-24 but still above historical norms. Recent work also documents diversification away from traditional reserve currencies, greater central bank demand for gold,  and increased attention to where gold is held (Arslanalp et al. 2026, ECB 2026). Yet the dollar remains central to global trade, finance, and the transmission of US monetary policy (Acharya et al. 2025). The policy debate is therefore often framed too starkly as ‘de-dollarisation versus dollar dominance’. For central banks, the more practical question is how reserve composition affects resilience when the Federal Reserve surprises markets.

Figure 1 illustrates the apparent tension. The dollar remains the largest reserve currency, even as its share of allocated foreign-exchange reserves has trended down and gold’s share of total reserves has recovered from its post-Global Crisis trough.

Figure 1 Composition of international reserves

Note: Panel (a) reports the composition of allocated official foreign exchange reserves. Panel (b) reports the market-valued gold share of total official reserves. Data on the Chinese yuan’s share are available from 2016 onwards.
Source: IMF COFER Database, IMF International Financial Statistics, Arslanalp et al. (2023), and World Bank.

US monetary tightening can travel rapidly across borders. Higher dollar returns induce portfolio rebalancing, tighten global financial conditions, and often appreciate the dollar. These effects are especially costly where firms, banks, or governments have dollar liabilities, since a weaker domestic currency raises the local-currency value of their debts. Recent contributions emphasise that the size of these spillovers depends on domestic credibility, foreign-currency borrowing, and access to official liquidity backstops (Kalemli-Ozcan and Unsal 2026, Curcuru and Martin 2026). But less is known about whether the composition of a central bank’s own reserve balance sheet changes the immediate exchange-rate response to the same US shock.

A high-frequency test

In new research, we examine this question using 108 Federal Open Market Committee (FOMC) announcements from 2009 to 2023 and minute-by-minute exchange rates for 18 advanced and emerging economies (Aizenman et al. 2026). We measure monetary policy surprises from changes in federal funds futures in a narrow 30-minute window around each announcement, following Bauer and Swanson (2023). We then ask whether the exchange-rate response differs with countries’ pre-existing holdings of total foreign-exchange reserves, dollar reserves, non-dollar reserves, and gold.

This design has a simple advantage. Reserve holdings are slow-moving and potentially endogenous to a country’s broader vulnerabilities. By contrast, an unexpected FOMC decision arrives at a precise time and is common to all countries. Observing exchange rates minute by minute allows us to compare how different reserve balance sheets condition the response to the same external shock, while reducing the risk that unrelated domestic news drives the result.

The first result confirms the force of US spillovers. A surprise ten-basis-point tightening is followed by an average depreciation of foreign currencies of about 0.4% within 20 minutes. There is little evidence of movement before the announcement.

The second result is that reserve composition matters. A country with total foreign-exchange reserves one standard deviation above the sample mean experiences about 0.04 percentage points less depreciation after the same shock – roughly one-tenth of the average response. The difference is larger for dollar reserves: countries with substantially larger dollar buffers experience up to 0.1 percentage points less depreciation. Comparable holdings of non-dollar reserves do not show the same clear relationship.

This pattern is consistent with a dollar-specific insurance mechanism. Dollar reserves can be sold, lent, pledged, or used for intervention without first being converted. They therefore provide the closest balance-sheet hedge against a shock that raises the global demand for dollars.

Gold is a complement, not a replacement

Gold reserves are also associated with smaller exchange-rate responses. A one-standard-deviation increase in gold reserves is linked to about 0.04 percentage points less depreciation, although the estimate is less precise than for foreign-exchange reserves. Gold is not as liquid as a dollar deposit or Treasury security, and mobilising it may require a sale, swap, or collateralised transaction. But it can still strengthen the central bank’s balance sheet through collateral value, valuation gains, and convertibility into dollar liquidity.

This distinction helps reconcile two developments that can otherwise appear contradictory. Central banks may diversify towards gold because it is not another country’s liability and may be less exposed to sanctions or custodial risk. At the same time, they still need dollar assets because the underlying shock, liabilities, and market-funding needs remain dollar-centred. Gold can improve robustness at the margin, but it does not eliminate the operational value of immediately usable dollar liquidity.

Reserve stocks can matter even when no intervention is observed in the event window. A large and liquid portfolio makes future dollar provision more credible. Market participants may therefore price not only actual reserve sales, but also the capacity to intervene, lend dollars domestically, or meet collateral calls. In this sense, reserves operate partly through deterrence and signalling.

The role of swap and repo lines

The strongest evidence for this interpretation comes from official dollar-liquidity arrangements. The Federal Reserve’s swap lines and Foreign and International Monetary Authorities repo facility are designed as backstops during stress, and earlier research shows that access reduces offshore dollar-funding strains (Ravazzolo and Goldberg 2022).

In our estimates, dollar and gold reserves matter most for countries without access to US swap or repo facilities. In that group, the exchange-rate attenuation associated with larger dollar reserves is similar in magnitude to the average response to the monetary surprise itself, while the gold-reserve association is also larger than in the full sample. By contrast, among countries with access to these facilities, reserve stocks have little additional explanatory power for the immediate exchange-rate response.

Figure 2 Reserve buffers and access to official dollar liquidity

Note: Estimated exchange-rate responses associated with a one-standard-deviation difference in dollar and gold reserves following a ten-basis-point increase in the federal funds rate. The top panels show countries without US swap or repo facilities; the bottom panels show countries with them. Shaded areas denote 95% confidence intervals.

The natural interpretation is that self-insurance and official backstops are partial substitutes. When a credible public dollar facility exists, markets need not rely as heavily on the country’s own reserve portfolio. Where access is absent or uncertain, the central bank’s balance sheet becomes more important. This result also cautions against treating global liquidity facilities as universal: access is selective and contingent.

Matching reserves to vulnerabilities

The reserve buffer is strongest where it should be most valuable: in economies with high dollar exposure. A ten-percentage-point increase in the share of external liabilities denominated in dollars amplifies the depreciation following a ten-basis-point US tightening by as much as 0.2 percentage points. In countries whose dollar liabilities are above the sample median, larger dollar and gold reserve holdings are associated with substantially smaller exchange-rate responses.

This links two balance sheets. The private and public sectors may be vulnerable because they owe dollars, while the central bank can partly offset that vulnerability by holding assets capable of supplying dollars or collateral under stress. Reserve adequacy should therefore not be reduced to one headline ratio. The currency, liquidity, custody, and legal accessibility of reserve assets should be assessed against the currency and maturity structure of external obligations.

Three policy implications follow. First, diversification should be organised in tranches. The liquidity tranche must remain immediately usable in the currencies in which stress is likely to occur; a broader investment or geopolitical-insurance tranche can hold gold and other assets. Second, swap and repo lines are valuable components of the global safety net, but they cannot replace sound domestic funding structures, credible policy frameworks, or prudent limits on unhedged foreign-currency debt. Third, gold’s usefulness should not be overstated: it has no contractual yield, its price is volatile, storage and mobilisation are costly, and repatriating it can increase security while reducing its financial utility.

Our evidence concerns the immediate exchange-rate response to identified US monetary surprises. Annual reserve data do not reveal which assets were actively deployed at the time, and access to liquidity lines is not randomly assigned. The findings should therefore be read as evidence that reserve balance-sheet capacity conditions spillovers, not as a precise estimate of the effect of a particular intervention.

Geopolitical fragmentation is changing the optimal reserve portfolio, but it has not removed the dollar’s central role in global finance. The relevant choice is not ‘dollars or gold’. It is how to combine immediately usable dollar liquidity, credible official backstops, and politically robust collateral so that a country’s reserve assets match its external vulnerabilities.

References

Acharya, S, O Akinci, S Miranda-Agrippino and P Pesenti (2025), “Monetary policy spillovers and the role of the dollar”, VoxEU.org, 9 April.

Aizenman, J, J Saadaoui, G S Uddin and N Yago (2026), “US Monetary Spillovers, Foreign Exchange, and Gold Reserves at Times of Geopolitical Fragmentation”, NBER Working Paper 35337.

Arslanalp, S, B Eichengreen and C Simpson-Bell (2023), “Gold as international reserves: A barbarous relic no more?”, Journal of International Economics 145, 103822.

Arslanalp, S, B Eichengreen and C Simpson-Bell (2026), “Our underappreciated international reserve system”, VoxEU.org, 21 March.

Bahaj, S and R Reis (2022), “Central bank swap lines: Evidence on the effects of the lender of last resort”, Review of Economic Studies 89(4): 1654-1693.

Bauer, M D and E T Swanson (2023), “An alternative explanation for the ‘Fed information effect’”, American Economic Review 113(3): 664-700.

Curcuru, S E and A Martin (2026), “Foreign currency funding risk and global financial stability”, VoxEU.org, 17 May.

European Central Bank (2026), “The International Role of the Euro”, June.

Gopinath, G, E Boz, C Casas, F J Díez, P-O Gourinchas and M Plagborg-Møller (2020), “Dominant currency paradigm”, American Economic Review 110(3): 677-719.

Kalemli-Ozcan, S and F Unsal (2026), “Global shocks are back: Emerging markets holding up”, VoxEU.org, 14 April.

Miranda-Agrippino, S and H Rey (2020), “US monetary policy and the global financial cycle”, Review of Economic Studies 87(6): 2754-2776.

Ravazzolo, F and L S Goldberg (2022), “The Fed’s international dollar liquidity facilities: Roles and effects”, VoxEU.org, 22 August.