The US dollar’s share of globally allocated foreign exchange reserves has fallen from over 70% in the late 1990s to approximately 58% today (Figure 1). This development pulls down the share of ‘Big 4’ currencies – inclusive of the dollar, euro, yen, and pound. For many observers, this headline decline is presented as a leading indicator of the dollar’s waning dominance in the international monetary system. For years, narratives have abounded about reasons for the pending decline, including but not limited to: the rise of the euro after 2000, China’s growing economic footprint, the IMF’s granting of reserve currency status to the Chinese renminbi in 2015, geopolitical fractures fuelled by the Russia-Ukraine war, the threat of financial sanctions, and changing institutional strength across countries.

Central banks hold official foreign exchange reserves (i.e. assets denominated in foreign currencies) to meet precautionary liquidity needs, service external debt, and stabilise their currencies during stress periods. Reserves are important at the country level and across the international monetary system (Arslanalp et al. 2026).

However, our research shows that an interpretation of the dollar’s global reserve status via changes in the dollar’s aggregate reserve share should be conducted with caution (Goldberg and Hannaoui 2026). Dollar share changes at the aggregate level conflate two fundamentally distinct channels: preferences for dollar assets as a share of reserves and changes in holdings of total reserves.

Figure 1 IMF currency composition of official foreign exchange reserves (COFER) currency shares (percentage)

Notes: Big 4 currencies include USD, JPY, EUR, and GBP. Currency Composition of Official Foreign Exchange Reserves (COFER) data covers 149 reporting countries and presents aggregated currency shares across total allocated reserves. The world share of allocated reserves is shown for Big 4 currencies and USD by the left axis, and for Non Big 4 currencies by the right axis. Source: Goldberg and Hannaoui (2026) construction using IMF COFER year-end values.

Where interpretations can go wrong

Instead of thinking about the dollar share of official reserves as just being about portfolio preferences for dollar assets, an exact decomposition separates different channels behind any change in the world dollar share of reserves between two time periods. We decompose the change in the global share of reserves held in US dollars into the following three components.

The first component reflects changes in countries’ portfolio preferences. It captures how much the global dollar share changes due to individual countries’ increase, or decrease, in the fraction of their reserves held in dollars, holding the size of their reserve portfolios fixed. As a consequence, countries with larger initial reserve holdings have a bigger influence on this component.

The second component reflects changes in the size of countries’ reserve portfolios, interacting with the difference between the dollar share of their portfolio and the average dollar share across all countries. This component captures how the global dollar share shifts when countries accumulate, or draw down, reserves, holding their portfolio preferences fixed. The key nuance here is that the reserve change effect on the aggregate depends on whether a country initially held a higher or lower portfolio allocation to dollar assets when compared with the global average. When countries with above-average dollar shares expand their reserves, the global dollar share tends to rise. Further, when they shrink, it tends to fall. The opposite holds for countries with below-average dollar portfolio shares.

The third component captures the interaction between these two forces. It reflects situations where countries simultaneously change both the size of their reserves and their portfolio preferences. This is typically a smaller second-order effect relative to the first two components.

One key takeaway from this decomposition is that changes in the global dollar share do not solely reflect shifts in countries’ preferences for dollar assets. Movements in the size of reserve portfolios alone — especially among large reserve holders whose portfolios differ from the global average — can push the global dollar share up or down, even if underlying preferences remain unchanged.

What do the data show (2015–2020)?

As the data on individual central bank portfolio allocations are not broadly published, our analysis relies on databases constructed by Ito and McCauley (2020) and Chinn et al. (2022), with supplemental information from Laser et al. (2024) and Arslanalp et al. (2022). This yields a sample of 72 countries for which portfolio composition data are available for 2015-2016, with 69 of these countries also having portfolio composition data for 2020. The 2015-2020 period coincides with a decline in Currency Composition of Official Foreign Exchange Reserves (COFER) dollar shares by 6.8 percentage points. This 72 country sample accounts for over 90% of allocated reserves in the 2015 COFER data.

Among the 69 countries in our sample for which we observe portfolio shares in both years, the net contribution of the portfolio preference channel to the aggregate dollar share change is essentially a wash. This near-zero aggregate contribution, however, masks substantial heterogeneity at the country level. Figure 2 tracks the number of countries increasing versus decreasing their dollar portfolio shares over rolling five-year windows. Across the 69 countries included, from the vantage point of 2020, roughly as many countries increased their dollar share as decreased it over the previous five years. Over this period, there is little evidence of systematic flight from the dollar.

Figure 2 Counts of countries by directional change in US dollar share over five-year windows

Notes: The sample presented above includes all countries with available US dollar share data in a given year and five years prior. Countries with five-year US dollar share changes smaller than 0.5 percentage points are classified as ‘No Change’. Source: Goldberg and Hannaoui (2026).

As large reserve countries play a disproportionate role in the aggregate, we find that a few countries may have played major roles in the currency composition of foreign exchange reserves. Russia, with an initial reserve balance of $309 billion and a 15.5 percentage point decline in its dollar share, contributes negatively through the portfolio preference channel. Switzerland, despite raising its dollar share over this period, exerts strong downward pressure on the global aggregate through the reserve quantity channel alone: its massive reserve accumulation of over $450 billion, from an initial base with a much lower-than-average dollar share, mechanically reduced the global aggregate. Switzerland’s contribution to the aggregate dollar share decline rivals or exceeds Russia’s. This is despite Switzerland moving toward the dollar, not away from it.

Using our decomposition, we provide insights also pertaining to the three large reserve holders missing 2020 portfolio data: China, India, and Nigeria. To estimate their unobserved contribution, we solve for their implied preference changes under alternative hypothetical scenarios for the world aggregate. Under our central scenario (a 6.8 percentage point aggregate decline), the implied dollar share reduction for this consolidated grouping is twice as large in aggregate terms as the collective effect of all other 69 countries.

Given the scale of China’s holdings, we conjecture that this effect is primarily driven by China. Indeed, there is a strong concentration of global foreign exchange reserves. China alone held over 25% of total official reserves in recent years, while the bottom 90% of countries together held less than a quarter. As a result, a handful of countries can dominate the constructed aggregate for reasons that may or may not reflect their appetite for dollars.

Traditional liquidity needs of central banks are the main drivers of dollar shares

To understand the underlying forces behind country-specific preference changes, we also estimate a panel regression model covering 75 countries from 1999 to 2023. We build on the established literature, which finds that exchange rate regime, currency composition of external debt, and bilateral trade patterns are the primary structural drivers of dollar portfolio shares. Countries that de facto peg to the dollar hold higher dollar shares; proximity to the euro area systematically lowers them; and higher dollar-denominated debt shares raise them.

We extend this common baseline in multiple directions. First, we introduce a measure of the investment tranche of official reserves. This measure captures the portion of holdings in excess of a country’s liquidity needs, proxied by short-term external debt coverage or three months of imports. The motivation for introducing this measure is the observation that reserve managers with a larger investment tranche have greater capacity to seek returns and diversify beyond the traditional safe-asset universe dominated by dollar Treasuries.

Second, we test the role of relative returns across currencies. Higher returns on nontraditional reserve currencies, which we build via a weighted composite of Australian, Canadian, and Korean sovereign yields, are associated with modest diversification away from dollars, but only through the investment tranche portion of portfolios. The role of the zero lower bound environment for US dollar rates, and of euro-dollar return differentials, is statistically detectable but quantitatively small in our analysis.

Additionally, we find that low geopolitical alignment with the US — measured by a country’s voting record in the United Nations General Assembly — does not, in general, imply a lower dollar share in official reserve portfolios. Indeed, controlling for other drivers, countries with lower UN voting alignment with the US tend to hold higher dollar shares.

The relationship is nuanced: we find evidence that geopolitical distance reduces dollar shares only once a country’s reserves are large enough to satisfy its foreign currency liquidity needs, as measured by its investment tranche. This suggests that geopolitical motivations for diversification can only operate after the primary precautionary function of reserves has been secured.

What drove the country changes in dollar share between 2015 and 2020?

Using Shapley-Owen values, we decompose the relative importance of the aforementioned drivers in increasing, or decreasing, dollar shares of official reserve portfolios across countries. The analytics show that changes in bilateral trade exposure to the US, and the dollar denomination of external debt, collectively account for nearly half of the explained variation in dollar shares in the weighted specification.

These macroeconomic fundamentals, not exclusively geopolitics, are the primary quantitative drivers. Geopolitical forces operate through the investment tranche and are concentrated in a small number of large, low-alignment reserve holders.

Takeaways and conclusions

Our findings suggest that relying on the headline COFER dollar share alone — without accounting for the concentration of reserve holdings or decomposing quantity and valuation effects — risks systematically misreading the dollar’s standing as the world’s global reserve currency. Specifically, shifts in the aggregates can reflect the reserve accumulation or draw-down decisions of a small number of large holders whose initial dollar shares happen to differ from the average — a mechanical consequence of concentration, not a signal of changing views on the dollar.

Additionally, and at least based on history, the drivers of portfolio allocations have continued to be traditional macroeconomic fundamentals: trade exposure to the US and euro area among the major reserve holders, and the currency composition of their external debt. For countries with large investment tranches — and particularly those with lower geopolitical alignment with the US — the capacity for discretionary diversification exists, but it operates at the margin and only after liquidity needs are satisfied.

Taken together, the evidence available through 2020 does not support a narrative of a broad, preference-driven retreat from the dollar. It does, however, point to a small number of large holders whose decisions will continue to shape the constructed global aggregates, whatever the underlying motivations for their changes.

Authors’ note: The views in this column are solely the responsibility of the authors and do not necessarily reflect the views of the Federal Reserve Bank of New York or the Federal Reserve System.

References

Arslanalp, S, B Eichengreen and C Simpson-Bell (2022), “The stealth erosion of dollar dominance and the rise of nontraditional reserve currencies”, Journal of International Economics 138, 103656.

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

Chinn, M D, H Ito and R N McCauley (2022), “Do central banks rebalance their currency shares?”, Journal of International Money and Finance 122, 102557.

Goldberg, L and O Hannaoui (2026), “Drivers of dollar share in official foreign exchange reserves”, NBER working paper 34888.

Ito, H and R N McCauley (2020), “Currency composition of foreign exchange reserves”, Journal of International Money and Finance 102, 102104.

Laser, F, A Mihailov and J Weidner (2024), “Currency compositions of international reserves: Recent developments”, Bank of Finland Institute for Emerging Economies Policy Brief 6: 1–8.