Zoekopties
Home Media Explainers Onderzoek & publicaties Statistieken Monetair beleid De euro Betalingsverkeer & markten Werken bij de ECB
Suggesties
Sorteren op
Roland Beck
Senior Team Lead - Economist · International & European Relations, International Policy Analysis
Vlad Burian
Georgios Georgiadis
Principal Economist · International & European Relations, International Policy Analysis
Peter McQuade
Senior Economist · International & European Relations, International Policy Analysis
Niet beschikbaar in het Nederlands

Geopolitics and foreign holdings of euro area government debt

By Roland Beck, Vlad Burian, Georgios Georgiadis and Peter McQuade

Published as part of the The international role of the euro, June 2025.

This special feature uses granular data from the ECB’s Securities Holdings Statistics (SHS) to investigate whether foreign holdings of euro area securities react to geopolitics. Where similar information is available from the US Treasury, a comparison is made with respect to foreign holdings of US Treasury securities. The analysis reveals several key insights. First, foreign investors hold almost a quarter of both euro area and US government debt. While these total foreign holdings are concentrated in countries geopolitically aligned with the West, foreign official sector holdings of euro area government debt are mainly held by non-aligned countries. Second, official foreign investors’ holdings of euro area government debt have remained generally resilient since Russia’s invasion of Ukraine. The decline in holdings of countries (excluding Russia) that are not geopolitically aligned with the West, which have dropped by 5% relative to pre-invasion levels, has so far been contained, highlighting the importance of upholding the rule of law. Econometric estimates suggest that geopolitical non-alignment explains at least part of this decline. These patterns still hold, even after correcting for geographic biases in international financial statistics to the extent possible. As the decline is small, the impact on euro area bond yields has been very limited thus far.

1 Motivation

The willingness of foreign investors to hold a country’s government bonds has important economic consequences. Greater appetite among foreign investors for a country’s government bonds lowers borrowing costs.[1] Since sovereign bond yields often serve as a benchmark for the pricing of other assets, this appetite also affects borrowing costs in other sectors. As such, changes in foreign holdings of government bonds can interfere with the transmission of monetary policy decisions to market interest rates. Foreign demand for a country’s government debt is also a key determinant of the international role of the country’s currency, influencing monetary policy spillovers and shaping the global monetary system.[2] Therefore, changes in foreign holdings of government debt are closely monitored by market participants, particularly for major reserve currencies such as the US dollar and the euro.[3]

Foreign holdings of euro area government debt are important from a monetary policy perspective. Overall, foreign holdings of government debt can have stabilising or destabilising effects on sovereign bond markets. Foreign investors may contribute to a more diversified investor base and lower bond yields.[4] For instance, as net purchases of euro area sovereign debt by the Eurosystem came to a halt at the end of June 2022, foreign holdings have contributed to the smooth absorption of sovereign debt issuances.[5] At the same time, foreign holdings could also have destabilising effects on sovereign bond markets in the event of sudden reversals of holdings, e.g. in response to global economic or political shocks. For example, during the euro area sovereign debt crisis, non-euro area investors appear to have “under-invested” in stressed euro area countries relative to the predictions of a gravity model.[6]

Geoeconomic risk in portfolio investment decisions has come to the fore since Russia’s invasion of Ukraine in 2022. The increase in geopolitical tensions over the past few years may have increased the importance of geopolitical factors as drivers of trade and financial flows. In the growing literature in this area, Gopinath et al. (2025) define “geoeconomic fragmentation” as “policy-induced changes in the sources and destinations of cross-border trade or financial flows guided by strategic considerations, such as national and economic security”.[7] The authors find some evidence for geoeconomic fragmentation in overall portfolio investment flows, while ECB research has also documented the importance of fragmentation for foreign direct investment.[8]

Against this background, this special feature explores the role of geoeconomic factors for holdings of euro area debt by foreign official sectors. Foreign official holders of government debt can be various types of entities including, most notably, central banks but also sovereign wealth funds and international organisations.[9] Since central banks typically hold foreign debt securities primarily for exchange rate intervention and management as well as other precautionary motives, the composition of their reserves tends to be determined by economic considerations. The latter may also include geoeconomic considerations, which can be economically rational if central banks anticipate that assets could become less liquid amid geopolitical tensions. In some cases non-economic, political motives can also play a role, for instance, for reserve managers and central banks with limited independence.

2 Stylised facts

To study foreign official holdings of euro area sovereign debt, this special feature uses, in addition to standard macroeconomic data, the third-party holdings of the ECB’s Securities Holdings Statistics (SHS). For euro area debt securities issued by general governments, the ECB’s data on the euro area international investment position (i.i.p.) provide information about the extent to which such holdings are held by non-residents (Table A.1). Analysing the geography of foreign holdings of government debt requires country-level information that is often not publicly available.[10] For the euro area, granular information on the holders of euro area government debt is therefore obtained from the ECB’s SHS data on “third-party holdings (TPH)”. While these data offer breakdowns by geography and sector for issuers and holders, they only cover holdings with euro area “custodians”, i.e. entities which are often a credit institution and provide securities custody services to their customers.[11] This means that the data may cover only part of the total non-euro area official holdings of euro area government debt, which may also be held via non-euro area custodians. At the same time, SHS data on non-resident holdings may also deviate somewhat from official i.i.p. data (Table A.1) as the latter are constructed based on the difference between reported intra-euro area assets and total euro area issuances, while SHS TPH data are based purely on custodial information.[12]

The US TICS provides information on total foreign holdings of US government debt, but a breakdown of private and official sector holders by holding area is not publicly available. As a result, distinguishing between foreign official and private sector holdings by holder country as in the ECB’s SHS data is not possible. In addition, there is a custodial bias in US TICS data, which do not look through holdings in European financial centres (Box A). Keeping this caveat in mind, the analysis starts by comparing total foreign holdings – that is, the sum of official and private holdings – of euro area and US government debt.

Table A.1

Sources and values of variables

Variable

Source

EUR bn

Euro area debt securities issued by general governments:

ECB, quarterly sector accounts

11,186.3

… of which held by non-residents

ECB, international investment position

2,591.5

ECB, Securities Holdings Statistics

2,674.2

… of which held by the foreign official sector

ECB, Securities Holdings Statistics

952.4

Sources: ECB, sectoral accounts and SHS.
Note: The latest observations are for the fourth quarter of 2024.

2.1 Overall foreign holdings

Over the last decade, foreign investors have been major holders of both euro area and US government debt, accounting for between one-fifth and one-third of all holdings of government debt in these economies. With around €2,600 billion at the end of the fourth quarter of 2024 (Table A.1), around 23% of outstanding euro area government debt was held by non-residents (Chart A.1). This share had started to decline when the Eurosystem acquired euro area government debt as part of its asset purchase programmes but has increased again since net purchases came to an end in mid-2022.[13] Holdings of US government debt exhibit a similar trend, as the share of non-resident holdings in total outstanding US Treasuries declined over the past decade and has recently risen to around 24%.[14]

Chart A.1

Foreign holdings of euro area and US government debt securities

a) Euro area government debt securities outstanding, by holder

b) US Treasury securities outstanding, by holder

Sources: ECB, US TICS and ECB staff calculations.
Notes: Euro area debt securities refer to securities issued by general governments. Data for the United States include all Treasury securities outstanding. The latest observations are for the fourth quarter of 2024.

To analyse foreign holdings in terms of geopolitical alignment, countries are assigned into three groups – aligned, non-aligned and connector countries. The split is based on estimates of bilateral geopolitical distance, reflecting voting patterns at the United Nations General Assembly, whereby “aligned countries” refer to countries in political proximity to the United States, “non-aligned countries” excluding Russia[15] refer to those in political proximity to China and “connector countries” serve as bridges between the hypothetical blocs.[16] While this measure is transparent and used in other empirical studies, some caveats should be kept in mind. For example, it might be an imprecise measure of geopolitical alignment as it does not account for differences in geopolitical relevance across the issues which countries are voting on and for strategic absences. In addition, geographic biases in international financial statistics could blur portfolio shifts between these country groups given the complex financial intermediation chains in financial centres (Box A). While these biases are corrected in the analysis to some extent, the evolution of geopolitical alignment patterns in foreign holdings should be treated with some caution.

Euro area government debt is largely held by geopolitically aligned countries, and increasingly so since 2022. According to information from third-party holdings in the ECB’s SHS,[17] aligned countries account for around 70% of foreign holdings of euro area government debt – a share which has increased somewhat since 2022 (Chart A.2, panel a).[18] Similarly, aligned-country holdings of US Treasuries account for around 55% of total foreign holdings, with a somewhat more significant role played by “connector countries” (Chart A.2, panel b).[19] As a result, the share of non-aligned countries in foreign holdings of government debt is around one-quarter in the euro area and around one-fifth in the United States.

Chart A.2

Foreign holdings of euro area and US government debt securities broken down by geopolitical blocs

a) Foreign holdings of euro area government debt securities, by geopolitical blocs

b) Foreign holdings of US Treasury securities, by geopolitical blocs

(EUR billion)

(USD billion)

Sources: ECB SHS, US Treasury and ECB staff calculations.
Notes: Euro area debt securities refer to securities issued by general governments excluding holdings of non-euro area EU countries. Data for the United States include all Treasury securities outstanding. The latest observations are for the fourth quarter of 2024.

Box A
Geographic biases in international financial statistics

Prepared by Roland Beck and Martin Schmitz

International financial statistics can be subject to geographic biases owing to complex financial intermediation chains. It can be particularly difficult to properly identify the holders of securities held in custody abroad as well as those held through investment funds. These biases could be particularly relevant in the analysis of geopolitical tensions, as financial flows may be redirected via financial centres to their final destinations.

With regard to the first bias, US TICS data do not look through custodians located outside the United States, e.g. in large financial centres, including from the euro area.[20] As a result, the rise in aligned holdings as shown in Chart A.2, panel b) could be biased if non-aligned investors hold securities through custodians located in these financial centres. One example that has been mentioned in this context are Chinese holdings of US Treasury securities through Euroclear in Belgium.[21]

With regards to foreign holdings of euro area government debt securities, the ECB’s SHS TPH data trace foreign holdings to the immediate counterpart of the reporting custodians (for custodians located in the euro area). The immediate counterpart may coincide with the actual holder in some cases (e.g. in the case of China where holdings are mainly held by the official sector).[22] However, SHS TPH data do not track holdings to the ultimate investor through chains of financial intermediation, e.g. in the case of holdings via global financial centres. For euro area government debt holdings, offshore financial centres as immediate counterparts account for only around 2% of overall foreign holdings of euro area debt and around 1.5% of foreign official holdings, having declined somewhat overall in recent years (Chart A.3).

Chart A.3

Holdings of euro area government debt securities by financial centres

a) Offshore financial centres’ holdings of euro area government debt securities

b) Offshore financial centres’ official holdings of euro area government debt securities

(left-hand scale: EUR billions, right-hand scale: percentages)

(left-hand scale: EUR billions, right-hand scale: percentages)

Sources: ECB SHS and ECB staff calculations.
Notes: The definition of offshore financial centres is based on the country list in Coppola et al. (2021) and includes data for the respective non-EU countries, such as, for example, the Cayman Islands, Hong Kong, Singapore and the Virgin Islands. The latest observation is for the fourth quarter of 2024.

In the case of investment funds, the economic ownership principle underlying international financial statistics implies that the assets of an investment fund are recorded as holdings of the country of residence of the investment fund. As a result, foreign holdings of government debt securities can be large for countries which serve as hubs in the investment fund industry. These countries often do not correspond to the countries where the holders of the underlying investment fund shares reside. For example, Luxembourg and Ireland are among the largest holders of Treasury securities according to US TICS data (Chart A.4, panel a), mainly due to their large investment fund industry. Estimates using security-level data suggest that euro area residents account for only around one-third (one-quarter in the case of bonds) of investment fund assets held by investment funds in Luxembourg and Ireland (Chart A.4, panel b).

Chart A.4

Geographic bias in foreign holdings of US Treasury securities

a) Top ten holders of Treasury securities

b) Security holdings of investment funds in Luxembourg and Ireland by type and residence of underlying owner

(USD billions)

(USD billions)

Sources: US TICS data and ECB staff calculations.
Notes: In panel a), the bars for the financial centres and Belgium (where Euroclear is located) are highlighted in yellow and red respectively. Panel b) reproduces figures reported in Beck, R., Coppola, A., Lewis, A., Maggiori, M., Schmitz M. and Schreger, J., “The geography of capital allocation in the euro area”, Working Paper Series, No. 3007, ECB, 2024. The authors provide an estimate of underlying euro area and rest of the world investor ownership of investment funds in Luxembourg and Ireland after performing a fund unwind procedure using security-level data. The latest observation is for the fourth quarter of 2024.

For around 40% of the investment fund shares issued in Luxembourg and Ireland, no owners can be identified in international financial statistics such as the IMF Coordinated Portfolio Investment Survey. However, there is some evidence that a part of these “missing assets” is held by investors from outside the euro area through custodians in the United Kingdom.[23] Hence, euro area investment fund holdings of US government securities as shown in Chart A.2, panel b) partly reflect the underlying investments of non-aligned countries.

2.2 Foreign official sector holdings

Geoeconomic alignment can be particularly relevant for official investors. Even when investing with economic objectives, official investors may be influenced by their government’s broader strategic considerations. In normal circumstances, central banks hold government debt for precautionary reasons and their holdings must meet certain safety and liquidity requirements,[24] while ideally also offering some return.[25] However, central banks might also internalise the risk that the liquidity of their holdings could become constrained when geopolitical tensions rise.

Euro area government debt held by the official sector in non-aligned countries has slightly decreased since Russia’s invasion of Ukraine, while total foreign official holdings rebounded to close to historic highs. Non-aligned countries excluding Russia, account for the bulk of foreign official sector holdings of euro area government debt (Chart A.5, panel a). Following Russia’s invasion of Ukraine in early 2022, non-aligned foreign official holdings dropped somewhat in the second quarter of 2022 and have increased slightly since then. From the end of 2021 onwards, the decline in these holdings has been limited to around 5% of pre-war holdings, highlighting the importance of upholding the rule of law. Moreover, they are still well above the levels witnessed at the time of Russia’s annexation of Crimea in 2014. After an initial drop in early 2022, foreign official holdings of aligned countries grew strongly, while those of connector countries also picked up slightly in the latter half of 2024. As a result, overall foreign official holdings rebounded, while the share of foreign official sector holdings held by non-aligned countries declined somewhat between the end of 2021 and the fourth quarter of 2024 (Chart A.5, panel b). While substantial as a share of foreign official sector holdings, non-aligned foreign official holdings accounted for a small portion, around 6%, of the outstanding euro area government debt in this period.

Chart A.5

Foreign official holdings of euro area government debt

a) Foreign official holdings of euro area government debt securities, by holder

b) Foreign official holdings of euro area government debt securities, by holder

(EUR billions)

(percentages of total foreign official holdings)

Sources: ECB SHS and ECB staff calculations.
Note: The latest observation is for the fourth quarter of 2024.

3 Empirical analysis

In addition to geopolitics, various other factors could drive the changes in foreign official sector holdings of euro area government debt. These could include, among others, valuation effects or changes in bond yields. To control for such factors, a standard gravity model is used to estimate the effect of Russia’s invasion of Ukraine on official sector holdings of non-aligned countries, adapting the analysis in Gopinath et al. (2025). The estimates are generated using bilateral data in a regression that is fully saturated with fixed effects at the country-pair, source-times and destination-time level to account for all the time-invariant, country-pair determinants of official securities holdings, as well as all time-varying source-specific and destination-specific factors.[26] The regressions track the change in the difference between aligned and non-aligned countries’ foreign official sector holdings of euro area government debt before and after Russia’s invasion of Ukraine. The regressions analyse the average difference in bilateral foreign official holdings between aligned and non-aligned countries excluding Russia, which can be termed the “geopolitical alignment differential”, following the invasion.

The analysis suggests that Russia’s invasion of Ukraine has increased the geopolitical alignment differential in foreign official sector holdings of euro area sovereign debt, although the estimates are imprecise and may not fully capture all factors driving shifts in official reserve portfolios. Official sector holdings of euro area sovereign debt by non-aligned countries, excluding Russia, are estimated to be almost 40% lower, on average, than holdings of aligned countries following the Russian invasion of Ukraine in 2022 (Chart A.6, panel a).[27] However, the size and significance of this finding is sensitive to the countries included in the non-aligned group and the estimate loses statistical significance at conventional levels if specific countries are excluded. Moreover, despite the comprehensive set of fixed effects, the estimates might not entirely account for other factors influencing changes in official reserve portfolios, as explained in greater detail below. The results also provide some evidence that there may be an effect on the holdings of connector countries, which are not assigned to either geopolitical bloc. Rather than an abrupt shift, the effect on non-aligned holdings appears to have materialised gradually, as shown by the estimates in Chart A.6, panel b).

Chart A.6

Estimated effect of Russia’s invasion of Ukraine on holdings of euro area sovereign debt in geopolitically distant countries

a) Average of non-aligned and connector countries in the post-invasion period

b) Non-aligned countries over time in the post-invasion period

(percentages)

(percentages)

Sources: ECB SHS and ECB staff calculations.
Notes: The diamonds show estimated percentage changes in the difference in holdings of euro area sovereign debt between more and less geopolitically distant countries after the Russian invasion of Ukraine. Estimates are obtained from regressions of a non-euro area foreign country’s official sector holdings of a euro area country’s government debt on dummy variables denoting whether the foreign country is in a different geopolitical grouping from the euro area country, interacted with a dummy variable that equals one after Russia’s invasion of Ukraine in the first quarter of 2022. Regressions are estimated on country pairs of 20 euro area issuers and 57 non-euro area foreign holder countries over the period from the first quarter of 2017 to the fourth quarter of 2024. Following Gopinath et al. (2024), the baseline specification includes country-pair fixed effects and source x time and host x time fixed effects to control for time-varying country-specific factors. The shaded areas indicate the 90% confidence intervals, based on standard errors clustered at the country-pair level. Panel b) shows the results of an analogous regression, but with the post-invasion variable and interactions replaced with individual dummy variables for each quarter after the Russian invasion of Ukraine.

The impact of changes in non-aligned official holdings on euro area bond yields is likely to have been contained thus far. Recent estimates of the link between portfolio changes and bond yields in the euro area suggest that a 10% purchase of government debt (around €1,120 billion) by the Eurosystem lowers euro area bond yields by 25 basis points.[28] Assuming these effects are symmetric and the same for all investor types, the decline in non-aligned holdings since the end of 2021 (i.e., €35 billion or 0.3% of euro area government debt) would lead to a negligible increase in euro area bond yields (less than 1 basis point).

These findings should be treated with caution as they are subject to several caveats. First, measuring geopolitical alignment is challenging. Using past United Nations voting behaviour is only one approach to measuring geopolitical alignment and the assessment is sensitive to some of the modelling choices. Second, the ECB SHS TPH data do not cover the full holdings of euro area securities by foreign investors, as the holdings of non-euro area custodians are not included. As a result, some euro area sovereign debt holdings in the reserves of non-aligned countries might have been transferred to custodians outside the euro area, which are not reflected in the SHS data. Third, this analysis is limited to debt securities holdings so that any eventual shifts of foreign reserves into other instruments such as deposits are not captured. Notably, in 2022 and 2023 policy interest rates in major economies rose significantly, prompting official reserve managers to mitigate duration risks by reallocating a portion of their reserves into deposits. Fourth, bilateral holdings data as used in this analysis are subject to custodial biases because the ultimate holders may not always correspond to the initial investors reported in the ECB SHS TPH. Finally, the estimates for the effect of foreign official purchases on euro area government bond yields are subject to several caveats as back-of-the-envelope calculations use elasticities borrowed from the literature on quantitative easing. These may not take into account geopolitical developments, which could lead to non-linearities in demand for international reserve assets and the related response of yields.

  1. See, for example, Rashad, A. and Rebucci, A., “Dollar Reserves and U.S. Yields: Identifying the Price Impact of Official Flows”, Journal of International Economics, Vol. 152, 2024.

  2. See Gräb, J. and Mehl, A., “Special feature A – The benefits and costs of the international role of the euro at 20”, The international role of the euro, ECB, 2019.

  3. See, for example, Goldberg, L. and Hannaoui, O.Z., “Drivers of Dollar Share in Foreign Exchange Reserves”, Federal Reserve Bank of New York Staff Reports, No 1087, March 2024. Revised August 2024.

  4. See, for example, Arslanalp, S. and Poghosyan, T., “Foreign Investor Flows and Sovereign Bond Yields in Advanced Economies”, Journal of Banking and Financial Economics, 2016, pp. 45-67.

  5. See “Box 1 – Sovereign bond markets and financial stability: examining the risk to absorption capacity”, Financial Stability Review, ECB, November 2023.

  6. See Beck, R., Georgiadis, G. and Gräb, J., “The geography of the great rebalancing in euro area bond markets during the sovereign debt crisis”, Journal of Empirical Finance, Vol. 38, Part A, 2016, pp. 449-460.

  7. See Gopinath, G., Gourinchas, P.-O., Presbitero, A.F. and Topalova, P., “Changing Global Linkages: A new Cold War?”, Journal of International Economics, Vol. 153, 2025.

  8. The authors use bilateral data on portfolio flows from the IMF’s Coordinated Portfolio Investment Survey (CPIS), including equity and debt securities while excluding cross-border direct investment and reserve assets. For evidence on foreign direct investment, see “Box 1 – Geopolitical fragmentation in global and euro area greenfield foreign direct investment”, Economic Bulletin, Issue 7, ECB, 2024.

  9. Sovereign wealth funds are more focused on the pursuit of return objectives compared with central banks, for which other considerations such as liquidity may be important. See Beck, R. and Fidora, M., “The Impact of Sovereign Wealth Funds on Global Financial Markets”, Review of European Economic Policy, Vol. 43, No 6, 2008 and Palacios, M.D. and Habib, M.M., “Box 1 – Sovereign wealth funds and the euro area: preliminary evidence”, The international role of the euro, ECB, 2024.

  10. The euro area’s international investment position provides a geographic breakdown by holding area, but only for total portfolio investment debt liabilities, not broken down by issuing sector.

  11. SHS data are collected on a security-by-security level (based on Regulation ECB/2012/24). For foreign investors’ third-party holdings data, the reported shares of global financial centres such as the United Kingdom and the United States, which show in the data as the primary non-euro area investors, may be exaggerated because of the presence of custodians in London and New York. For more detailed information on SHS data, see the dedicated section of the ECB’s website as well as “Who holds what? New information on securities holdings”, Economic Bulletin, Issue 2, ECB, 2015.

  12. As a result, some of the holdings which TPH data show as belonging to non-euro area investors may actually reflect euro area holdings.

  13. See Koijen, R.S.J., Koulischer, F., Nguyen, B. and Yogo, M., ”Euro-Area Quantitative Easing and Portfolio Rebalancing”, American Economic Review, Vol. 107, No 5, 2017, pp. 621-27; Bergant, K., Fidora, M. and Schmitz, M., “International capital flows at the security level: evidence from the ECB’s Asset Purchase Programme”, Working Paper Series, No 2388, ECB, 2020.

  14. In its reports about foreign holdings of US Treasury securities, the Congressional Research Service refers to the share of foreign holdings in total marketable Treasury securities held by the public, excluding holdings held by the Federal Reserve System. As the total outstanding amount of Treasuries is therefore reduced, the reported share of foreign ownership stands at 31% as at the end of 2023.

  15. To ensure that results are not driven by mechanical effects, the sample excludes Russia since the assets of the Bank of Russia held in the EU were immobilised in response to Russia’s invasion of Ukraine.

  16. See Gopinath et al. (2025) and the references provided therein.

  17. Total foreign holdings of euro area government debt roughly match the information in the euro area’s international investment position (Table A.1).

  18. Following Gopinath et al. (2024), countries are divided into three groups based on the 2021 ideal points distance. The aligned bloc includes countries in the top quartile in their political proximity to the United States, while the non-aligned bloc includes countries in the top quartile in their political proximity to China. The connector countries bloc comprises the remaining economies.

  19. US TICS data are corrected for Chinese holdings of US Treasury securities via Euroclear in Belgium using as an upper bound the assumption that all Treasury holdings in Belgium originate from China, in the spirit of Setser (2023), op. cit.

  20. According to the US Treasury, “this custodial bias contributes to the large recorded foreign holdings of U.S. securities in major financial centers, such as Belgium, the Caribbean banking centers, Luxembourg, Switzerland, and the United Kingdom.”

  21. See Setser, B.W., “China Isn't Shifting Away From the Dollar or Dollar Bonds”, Council on Foreign Relations, Blog Post, October 2023.

  22. See Bosetti, I., Incardona, R. and Rodríguez Caloca, A., “Filling the gap: the geographical allocation of euro area portfolio investment liabilities and related income”, Statistics Paper Series, No 50, ECB, 2025.

  23. See Beck, R., Coppola, A., Lewis, A., Maggiori, M., Schmitz, M. and Schreger, J., “The geography of capital allocation in the euro area”, Working Paper Series, No 3007, ECB, 2024 and Milesi-Ferretti, G.M., “Many Creditors, One Large Debtor: Understanding The Buildup of Global Stock Imbalances after the Global Financial Crisis,” IMF Economic Review, 2023.

  24. Central banks may also hold reserves for investment purposes and sometimes split their reserve portfolio into a liquidity tranche and an investment tranche. In emerging economies, central banks may also hold foreign reserves to cushion shocks to the availability of foreign financing for imports and external debt payments. For a more detailed review of the motives for holding foreign exchange reserves, see, for example, “Assessing Reserve Adequacy—Specific Proposals”, IMF Policy Paper, IMF, Washington, 2015.

  25. See, for example, “HSBC Reserve Management Trends 2023”, edited by Robert Pringle and Joasia E. Popowicz, April 2024.

  26. The reported regressions include non-EU financial centre holdings, but the results remain largely unchanged if these holdings are excluded in the regression framework.

  27. The coefficient estimate is slightly larger if Russia is included in the sample, likely reflecting the mechanical impact of immobilised assets of the Bank of Russia held in central securities depositories in the European Union.

  28. These estimates were obtained in the context of quantitative easing. See Koijen, R.S.J., Koulischer, F., Nguyen, B. and Yogo, M., “Inspecting the mechanism of quantitative easing in the euro area”, Journal of Financial Economics, Vol. 140, Issue 1, 2021, pp. 1-20.