Back Matter
  • 1 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 2 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 3 https://isni.org/isni/0000000404811396, International Monetary Fund

Annex 1. Existing Literature on Drivers of Reserve Currency Shares

The literature examining the drivers of reserve currency shares at the global level finds a significant role for the economic characteristics of reserve issuers, such as their global reach (generally captured by their economic size and/ or role in international trade and finance, also aiming to capture “network effects”)1 and credibility (Li and Liu 2008; Eichengreen, Mehl, and Chiţu 2016; Aizenman, Cheung, and Qiand 2019). Some studies also point to the role of national policies in either supporting or preventing the internationalization of currencies (Eichengreen, Mehl, and Chiţu 2016).2 Furthermore, some studies capture the reserve currencies’ perceived safety and effective medium of exchange by relating it to the depth and liquidity of onshore and offshore financial markets (Chinn and Frankel 2008).3

In addition to global reach, network effects, and credibility, the literature finds that inertia plays an important role; that is, holding a larger share of a given reserve currency in the past tends to be a good predictor of reserve shares in the future, as discussed by Triffin (1960). This suggests that reserve currency take-up may be nonlinear, with a high degree of inertial bias in favor of the incumbent reserve currency (Frankel 2012). As such, reserve currency choices may be informed less by short-term economic fundamentals and more by historical ties.4 The literature also concludes that, after the collapse of the Bretton Woods system, inertial effects became stronger, which may reflect the higher stability in the US dollar’s share after the shift from the pound to the dollar. By contrast, the network effects seem to be weaker post-Bretton Woods, which may reflect lower switching costs due to advances in financial and transactions technology (Eichengreen, Mehl, and Chiţu 2016).

A few studies argue that geopolitical considerations can also play a role, as countries may choose to hold reserves in a given currency because of geopolitical or strategic considerations, or as a result of military alliances, and so reserve currencies’ perceived safety can be linked to reserve issuers’ geopolitical or military power (Cohen 2015; Kindleberger 1970; Posen 2008; Liao and McDowell 2016). Eichengreen, Mehl, and Chiţu (2017) show that military alliances boosted the shares of the currencies of alliance partners in foreign reserve portfolios by close to 30 percent in the run up to World War I.

Studies using aggregate data, however, fail to account for shifts within individual countries’ portfolios and cannot capture reserve holders’ potential transactional demand (the intended uses of the reserves). Central banks’ reserve portfolio decisions are often influenced by the pattern of a country’s transactional demands, including the structure of its trade and financial payments, and foreign exchange arrangements.5 More specifically, the trade links can be captured by the share of trade with the reserve currency issuer in the absence of granular data on trade invoicing, financial links by the currency composition of public debt or cross-border bank claims, and the foreign exchange market intervention by de facto anchoring to a reserve currency.

The literature using individual country data is relatively sparse due to lack of publicly available data. Studies using confidential COFER country-level data find evidence that the reserve holders’ potential transactional demand for trade and finance-related payments and foreign exchange market intervention drives the currency composition of their reserves (Heller and Knight 1978; Dooley, Lizondo, and Mathieson 1989; Eichengreen and Mathieson 2000).6

More specifically, Heller and Knight (1978), using data for 55 countries during 1970–76, find that the exchange rate regime (choice of peg) and trade linkages with reserve issuers matter. In turn, Dooley, Lizondo, and Mathieson (1989) show that the currency denomination of debt service payments is also a significant driver. Eichengreen and Mathieson (2000) highlight the stability of the currency composition of reserves over time and in relation to its main determinants (exchange rate links and trade and financial flows) during 1971–95, and find evidence that capital account liberalization in emerging market economies raises the share of currencies from reserve issuers with particularly active financial markets (United States and United Kingdom).

More recent work has focused on the links between reserve currencies and the currencies used for trade invoicing and financial claim denomination. Gopinath (2015), Gopinath and Stein (2018), and Gopinath and others (2020) emphasize the dominance of the US dollar and euro in trade invoicing, beyond direct trade links with the United States and the euro area.7,8 Ito, McCauley, and Chan (2015) and Ito and McCauley (2019) also highlight the role of the trade invoicing and financial liabilities denomination, as well as exchange rate comovements with reserve currencies, using publicly available country-level data.

Annex 2. Drivers of Reserve Currencies using Aggregate Data

This annex analyzes the drivers of reserve currencies’ shares in global reserve holdings using aggregate data since 1947. In line with the existing literature, the authors find that inertia and the credibility of reserve currency issuers are significant drivers of reserve configurations. In contrast to previous studies, the authors find evidence of a limited role for reserve issuers’ trade and financial reach, and, after controlling for these factors, no role of their geopolitical reach.

Empirical Specification

Building upon previous literature, the authors investigate global reserve currency shares using data sourced from Eichengreen, Mehl, and Chiţu (2016) before 1995 and COFER since 1995, and covering the period 1947–2018. The core specification, in line with the existing literature, considers the three factors typically found to be important drivers of aggregate reserve shares: inertia, global reach/size, and credibility.1 Specifically, the aggregate reserve share of currency i in year t is modeled as:

Reserve Sharei,t = αi + δt + Reserve Sharei,t-1 + GDP Sharei,t-1 + Credibilityi,t-1 + εi,t

in which:

  • αi is a currency random/fixed effect

  • δt is a time fixed effect

  • GDP Share is the share of the reserve issuer’s economy in global GDP

  • Credibility is the average appreciation of the reserve issuer’s currency against the SDR in the previous five years.

The authors use different econometric specifications and sample periods. The baseline specification uses currency fixed effects, since the Hausman test rejects random effects and the Arellano-Bond (1991) estimator has limitations due to the small cross section relative to the time horizon (Arellano 2003).

The authors also split the sample into 1947–98 and 1999–2018 to assess whether the importance of different drivers has changed since the introduction of the euro. The Deutsche mark, the French franc, and the Dutch guilder are used prior to 1999, and the euro is used starting in 1999.2

Results

Results under the baseline specification highlight the importance of inertia and credibility (Annex Table 1)3:

  • Inertia: Consistent with previous literature, inertia effects are large and significant across econometric specifications. The coefficients on lagged reserve shares are about 0.9, indicating a high degree of persistence.

  • Credibility: Coefficients are significant over the full sample period, but the economic effect is more limited, with a 10 percent appreciation of a currency associated with a 0.4 percentage point increase in its share of global reserves.

  • Size: In contrast to previous literature, the relative size of the economy of the reserve currency issuer, measured by its share of global GDP, is not robustly significant across specifications. Although the coefficient on size is positive and significant in the random effects specification, consistent with Eichengreen, Mehl, and Chiţu (2016), the sign is negative in other specifications and insignificant when using fixed effects.

  • Time variation: Results (Annex Table 1, columns 4 and 5) indicate that inertia and credibility effects may have been more important in the earlier period (1947–98).

Annex Table 1.

Baseline Specification

article image
Source: Authors’ calculations.Note: Robust standard errors in parentheses. ***, **, * indicates significance at the 1%, 5%, and 10% levels, respectively. All specifications include time fixed effects. Reserve shares are sourced from Eichengreen, Mehl, and Chiţu (2017) for 1947–2013 and the Currency Composition of Official Foreign Exchange Reserves database for 2014–18. GDP data is sourced from the Maddison Project database for 1947–2016, with data for 2017 calculated using IMF data on GDP based on PPP. “Credibility” is the average appreciation of the reserve currency against the SDR in the previous five years. “GDP” is the share of world GDP, which the reserve currency issuer accounts for. “USD” is a dummy variable equal to one if the reserve currency is the US dollar and zero otherwise. “No of groups” refers to the number of reserve currencies included. PPP = purchasing power parity; SDR = special drawing rights.

Robustness Checks

The authors test for heterogeneity in coefficients for specific currencies and whether longer lags matter. In particular, the authors consider whether the effects of credibility and size vary for the US dollar (Annex Table 1, column 6), which is the currency with the largest share of reserves throughout the sample period. Results suggest that credibility is not an important factor for the US dollar share of aggregate reserves. This may indicate that once a reserve currency is widely used, short-term episodes of depreciation are less important. Alternatively, periods when the US dollar is appreciating may reflect flight to safety effects, rather than the underlying credibility of the United States as a reserve issuer. The authors also include longer lags of reserve shares and credibility and size measures, but these generally are not statistically significant and do not materially change the results.

The coefficient for the reserve currency issuer’s share of global GDP is only positive and significant in the random effects specification, suggesting that GDP shares could be a poor proxy for the global reach of reserve issuers and their importance in global trade and financial networks. Instead, direct measures of trade and financial reach are considered as alternative measures of global reach:

To measure the importance of a reserve currency issuer in global trade networks,the country’s share of world exports and a measure of its centrality in the global trade network are used (see Papamichalis and others, forthcoming).4 While the estimated coefficients are positive, they are not significant at the 5 percent level (Annex Table 2, column 3). Thus, the authors find weak evidence that prominence in global trade networks matters for reserve currency shares.

Annex Table 2.

Alternative Measures of Global Reach: Trade and Financial Reach

article image
Source: Authors’ calculations.Note: Robust standard errors in parentheses. ***, **, * indicates significance at the 1%, 5%, and 10% levels, respectively. All specifications include currency and time fixed effects. For comparison, column 1 shows the baseline specification presented in Annex Table 1. “Export share” is the share of the reserve currency issuer’s exports in world exports, using the Direction of Trade Statistics. “Export Centrality” is a measure of eigenvector centrality of the reserve currency issuer in the world trade network, calculated using the Direction of Trade Statistics. “FX Turnover” is the currency share of turnover in over-the-counter foreign exchange markets (BIS Triennial Central Bank Survey). “Debt Securities” is the share of long-term external debt of low- and middle-income countries in a given currency share of the outstanding stock of long-term external debt for low and middle-income countries (World Bank. International Debt Statistics). “Bank claims” is the currency share of total cross-border bank claims excluding unallocated currencies (Bank for International Statistics, Locational Banking Statistics). “No of groups” refers to the number of reserve currencies included.

The currency denomination of external debt in low- and middle-income countries matters for global reserves shares (Annex Table 2, column 5). Other measures of financial development—the share of foreign exchange turnover in a given currency and the share of cross-border bank claims in a given currency— are not found to be significant drivers of reserve shares.5 A possible explanation: reserve currency issuers typically tend to be highly financially developed and their currencies have usually achieved international status, so incremental gains in this context may not be important for reserve currency shares.

Eichengreen, Mehl, and Chiţu (2017) suggest that reserve currency choices may be influenced by geopolitical interests, with countries choosing to hold reserves of issuers which have diplomatic or military power. Since military alliances display little variation during the sample period, four alternative measures of geopolitical influence are considered: GDP per capita relative to the average GDP per capita of reserve issuers; the proportion of countries that have voted in the same direction as the reserve issuer country at the UN General Assembly in a given year using the data set detailed in Voeten, Strezhnev, and Bailey (2009)6; spending on official development assistance (ODA) as a share of GDP using OECD data; and military spending as a share of total military spending by reserve issuers using Stockholm International Peace Research Institute (SIPRI) data. The authors find no evidence that geopolitical factors have driven aggregate reserve shares during the sample period, once other factors are controlled for (Annex Table 3)—the sign on the geopolitical variables is negative for three of the measures and is insignificant for all measures. For instance, the proportion of countries voting in line with the United States has fallen sharply while the dollar’s share of reserves has been resilient (Annex Figure 1).7 Whether the measures used are poor proxies for geopolitical considerations, or whether political considerations are less relevant than they have been historically remain open questions.

Annex Table 3.

Alternative Measures of Global Reach: Geopolitics

article image
Source: Authors’ calculations.Note: Robust standard errors in parentheses. ***, **, * indicates significance at the 1%, 5%, and 10% levels, respectively. All specifications include currency and time fixed effects. For comparison, column 1 shows the baseline specification presented in Annex Table 1. GDP per capita data is sourced from the Maddison Project dataset. “UN Votes” is the share of votes by all countries at the UN General Assembly which have been in agreement with the votes of the reserve currency issuing countries, where abstentions are counted as “no” votes (Voeten 2013). “ODA” is the amount of official development assistance provided by the reserve currency issuer country as a share of the currency issuer’s GDP (Organisation for Economic Co-operation and Development). “Military Spending” is the military expenditure of the reserve currency issuer country as a share of total military spending by all reserve currency issuers (Stockholm International Peace Research Institute). “No of groups” refers to the number of reserve currencies included.
Annex Figure 1.
Annex Figure 1.

UN Voting Agreement and Reserve Currency Shares

Citation: Departmental Papers 2020, 002; 10.5089/9781513560298.087.A999

Sources: Voeten, Strezhnev, and Bailey (2009); and IMF staff calculations.

Annex 3. Drivers of Reserve Currencies using Disaggregated Data

Disaggregated data for 42 economies covering the period of 1999–2018 are used to further investigate the drivers of reserve currency shares by using more granular trade, financial, and geopolitical measures linking reserve holders to reserve currencies and their issuers. The authors find that inertia effects remain a key driver of reserve currency shares, although of smaller magnitude than in the aggregate data, while financial considerations are important but vary over time and across regions.

Empirical Specification

In line with previous literature, the baseline specification focuses on reserve holders’ potential transactional demands (the intended uses of the reserves) and bilateral links to reserve issuers as drivers of reserve currency shares. More specifically, the trade share with the reserve currency issuer proxies for trade links, the currency denomination of public debt or cross-border bank claims captures financial considerations, while the de facto anchoring to a reserve currency captures exchange rate stability considerations for intervening in the foreign exchange market.1 The authors use a panel of 42 economies with data available for some or all years from 1999 to 2018 for reserve holdings in the main four reserve currencies (US dollar, euro, Japanese yen, and British pound),2 and model the reserve share of currency i in country c’s reserve portfolio in year t as:

Reserve Sharec,i,t = αc,i + δt + Reserve Sharec,i,t-1 + Trade Sharec,i,t + FX Alignmentc,i,t + Financial Linksc,i,t + εi,t

where:

  • αc,i is a country-currency random/fixed effect

  • δt is a time fixed effect

  • Trade Sharec,i is the share of country c’s trade with reserve issuer i

  • FX Alignmentc,i is the estimated country c’s exchange rate comovement with the reserve currency i , following Ilzetzki, Reinhart, and Rogoff (2019)

  • Financial Linksc,i is either the share of country c’s public debt or cross-border bank claims denominated in reserve currency i.3

The baseline specification uses a model with country-currency and year fixed effects, as the Hausman test rejects the random effects model and the Arellano-Bond estimator has limitations due to the small sample size. A Tobit model addressing the fractional nature of the dependent variable provides qualitatively similar results.4

The authors estimate their model separately for AEs and EMDEs due to different drivers of reserve holdings and also different data availability across the two sets of countries. For example, debt considerations are much more relevant for EMDEs than for AEs, while the currency denomination of cross-border bank claims is available for AEs but not for EMDEs. The authors use the 2019 World Economic Outlook (WEO) classification to split the sample into EMDEs (32) and AEs (10, of which 6 are European). The findings do not change if instead the 2001 WEO or the contemporaneous WEO country classifications is used.

Results

Results under baseline specification highlight the importance of inertia and financial considerations (Annex Table 4):

  • Inertia: The findings suggest relatively large and significant inertia effects although of smaller magnitude than in aggregate data, with coefficients on the lagged reserve currency share in a range of 0.6–0.7 for EMDEs and 0.7–0.8 for AEs.

  • Financial considerations: The currency denomination of public debt (EMDEs) and cross-border bank claims (AEs) are also statistically significant, although economically less significant than inertia. A 10 percent increase in the share of a given currency denomination in public debt or bank claims is associated with about 1 percentage points increase in that currency’s reserve share.

  • Trade links and FX alignment: The measures of trade links and exchange rate comovement are not statistically significant determinants of reserve currency shares.

Annex Table 4.

Econometric Specifications

article image
Source: Authors’ calculations.Note: Robust standard errors in parentheses. ***, **, * indicates significance at the 1%, 5%, and 10% levels, respectively. All specifications include time and country-currency fixed (random) effects. Reserve shares for four reserve currencies (US dollar, British pound, Japanese yen, and euro) are obtained from the central banks’ websites. “Trade Share” is the share of trade with the reserve issuer, obtained from the Direction of Trade Statistics. “FX Alignment” is a dummy variable equal to one if the exchange rate co-moves with the reserve currency, as estimated by Ilzetzki, Reinhart, and Rogoff (2019). “Debt Share” is the share of public debt denominated in a given reserve currency, obtained from the World Bank International Debt Statistics dataset. “Bank Claims Share” is the share of cross-border bank claims denominated in a given reserve currency obtained from the BIS international banking data by location. AE 5 advanced economies; EMDE 5 emerging market and developing economies.

The importance of financial considerations has increased since the GFC (Annex Table 5). Also, the currency denomination of public debt is a significant determinant of reserve holdings in EMDEs in all regions except Middle East and Central Asia, and economically very significant in Africa and Asia-Pacific, while the currency denomination of cross-border bank claims is only statistically significant in AEs in Europe.

Annex Table 5.

Alternative Region and Time Periods

article image
Source: Authors’ calculations.Notes Robust standard errors in parentheses. ***, **, * indicates significance at the 1%, 5%, and 10% levels, respectively. All specifications include time and country-currency fixed effects. Reserve shares for four reserve currencies (US dollar, British pound, Japanese yen, and euro) are obtained from the central banks’ websites. “Trade Share” is the share of trade with the reserve issuer, obtained from the Direction of Trade Statistics. “FX Alignment” is a dummy variable equal to one if the exchange rate co-moves with the reserve currency, as estimated by Ilzetzki, Reinhart, and Rogoff (2019). “Debt Share” is the share of public debt denominated in a given reserve currency, obtained from the World Bank International Debt Statistics dataset. “Bank Claims Share” is the share of cross-border bank claims denominated in a given reserve currency obtained from the BIS international banking data by location. AE 5 advanced economies; EMDE 5 emerging market and developing economies.

After the GFC, inertia effects appear to have become stronger for both EMDEs and AEs (Annex Table 5, columns 1–2 and 8–9), with inertia effects for EMDEs converging with those for AEs in terms of magnitude. Inertia effects are largest among AEs in the Americas, and generally larger in AEs compared to EMDEs across all regions.

Trade links have become more important for AEs since the GFC (Annex Table 5, column 9). This is particularly striking for European countries, although not surprising given their trade links with Euro area economies.

Robustness Checks

Alternative measures of trade links and exchange rate comovement do not change the main results (Annex Table 6). The coefficient on trade links becomes positive but continues to be statistically insignificant when the share of trade is replaced by the share of imports, to account for the fact that for some countries the reserves are held mainly to cover purchases of foreign goods and services. Similarly, when the trade shares are replaced with currency shares in imports invoicing, obtained from Boz and others (2020), the coefficients continue to be imprecisely measured, possibly due to little variability in invoicing shares over time and hence collinearity with the fixed effects.

Annex Table 6.

Alternative Measures: Imports and Invoicing Share, De Jure Peg, Total Level of Reserves

article image
Source: Authors’ calculations.Note: Robust standard errors in parentheses. ***, **, * indicates significance at the 1%, 5%, and 10% levels, respectively. All specifications include time and country-currency fixed effects. Reserve shares for four reserve currencies (US dollar, British pound, Japanese yen, and euro) are obtained from the central banks’ websites. “Trade Share” is the share of imports with the reserve issuer, obtained from the Direction of Trade Statistics., or share in imports invoicing, obtained from Boz and others (2020). “FX Alignment” is a dummy variable equal to one if the exchange rate co-moves with the reserve currency, as estimated by Ilzetzki, Reinhart, and Rogoff (2019). In column (4), “FX Alignment” dummy is replaced with the de jure peg to a reserve currency from the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions database. “Debt Share” is the share of public debt denominated in a given reserve currency, obtained from the World Bank’s International Debt Statistics dataset. “Bank Claims Share” is the share of cross-border bank claims denominated in a given reserve currency obtained from the Bank for International Settlements international banking data by location. Data on the levels of total reserves come from IMF’s International Financial Statistics database. AE 5 advanced economies; EMDE 5 emerging market and developing economies.

The authors also find that trade links are a significant determinant of reserve shares in EMDEs with lower level of total reserves (measured by the log of total reserves from IMF IFS database), becoming less important as the level of reserves rises (Annex Table 6, column 5).5 Finally, the coefficient on anchoring turns positive, as expected, but remains statistically insignificant when using the measure of the de jure peg to a reserve currency from the IMF AREAER database.6

Geopolitical measures deliver mixed results (Annex Table 7). Tree measures of geopolitical influence are considered: the proportion of votes cast the same way as the reserve issuer at the UN General Assembly in a given year; the share of official development assistance (ODA) received from a given reserve issuer (for EMDEs only); and the share of imports of arms and ammunition (classified under Harmonized System Chapter 93) from a given reserve issuer. The results are mixed, and only the UN votes in AEs and the share of arms imports from reserve issuers in EMDEs are positively associated with the reserve currency share.

Annex Table 7.

Alternative Geopolitical Measures

article image
Source: Authors’ calculations.Note: Robust standard errors in parentheses. ***, **, * indicates significance at the 1%, 5%, and 10% levels, respectively. All specifications include time and country-currency fixed effects. Reserve shares for four reserve currencies (US dollar, British pound, Japanese yen, and euro) are obtained from the central banks’ websites. “Trade Share” is the share of trade with the reserve issuer, obtained from the Direction of Trade Statistics. “FX Alignment” is a dummy variable equal to one if the exchange rate co-moves with the reserve currency, as estimated by Ilzetzki, Reinhart, and Rogoff (2019). “Debt Share” is the share of public debt denominated in a given reserve currency, obtained from the World Bank’s International Debt Statistics dataset. “Bank Claims Share” is the share of cross-border bank claims denominated in a given reserve currency obtained from the Bank for International Settlement’s international banking data by location. “UN Votes” is the share of votes cast the same way as the reserve issuer at the UN General Assembly, with abstentions counted as “no” votes (Voeten 2013). “ODA Share” is the share of official development assistance received from the reserve issuer, sourced from the Organisation for Economic Co-operation and Development. “Share in Arms Imports” is the share of imports of arms and ammunition (classified under Harmonized System Chapter 93) from the reserve issuer, obtained from UN Comtrade. AE 5 advanced economies; EMDE 5 emerging market and developing economies; ODA 5 official development assistance.
Annex Table 8.

List of Economies

article image
Source: Authors’ compilations.

References

  • Aizenman, Joshua, Yin-Wong Cheung, and Xingwang Qiand. 2019. “The Currency Composition of International Reserves, Demand for International Reserves, and Global Safe Assets.” Journal of International Money and Finance 102 (C): 10220.

    • Search Google Scholar
    • Export Citation
  • Arellano, Manuel. 2003. Panel Data Econometrics. Oxford: Oxford University Press.

  • Arellano, Manuel, and Stephen Bond. 1991. “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations.” The Review of Economic Studies 58 (2): 27797.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bacchetta, Philippe, and Eric van Wincoop. 2005. “A Theory of the Currency Denomination of International Trade.” Journal of International Economics 67 (2): 295319.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bank of Canada (BoC), Bank of England (BoE), and Monetary Authority of Singapore (MAS). 2018. “Cross-Border Interbank Payments and Settlements: Emerging Opportunities for Digital Transformation.”

    • Search Google Scholar
    • Export Citation
  • Behar, Alberto, Kristina Kostial, and Rafael Ramírez. 2018. “Work on Long-Term Trends and Uncertainties at the IMF.” European Court of Auditors Journal 10: 3537.

    • Search Google Scholar
    • Export Citation
  • Bénétrix, Agustín, Deepali Gautam, Luciana Juvenal, and Martin Schmitz. 2019. “Cross-Border Currency Exposures. New Evidence Based on an Enhanced and Updated Dataset.” IMF Working Paper 19/299, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boar, Codruta, Henry Holden, and Amber Wadsworth. 2020. “Impending Arrival – A Sequel to the Survey on Central Bank Digital Currency.” BIS Papers 107, Bank for International Settlements, Basel.

    • Search Google Scholar
    • Export Citation
  • Boz, Emine, Camila Casas, Georgios Georgiadis, Gita Gopinath, Helena Le Mezo, Arnaud Mehl, and Tra Nguyen. 2020. “Patterns in Invoicing Currency in Global Trade.” IMF Working Paper 20/126, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brunnermeier, Markus, Harold James, and Jean-Pierre Landau. 2019. “The Digitalization of Money.” NBER Working Paper 26300, National Bureau of Economic Research, Cambridge, MA.

    • Search Google Scholar
    • Export Citation
  • Caballero, Ricardo J., Emmanuel Farhi, and Pierre-Olivier Gourinchas. 2017. “The Safe Assets Shortage Conundrum.” Journal of Economic Perspectives 31 (3): 2946.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Carney, Mark. 2019. “The Growing Challenges for Monetary Policy in the Current International Monetary and Financial System.” Speech at “Challenges for Monetary Policy,”, a symposium sponsored by the Federal Reserve Bank of Kansas City, Jackson Hole, WY, February 12.

    • Search Google Scholar
    • Export Citation
  • Chahrour, Ryan, and Rosen Valchev. 2017. “International Medium of Exchange: Privilege and Duty.” Boston College Working Papers in Economics 934, Boston College, Chestnut Hill, MA.

    • Search Google Scholar
    • Export Citation
  • Chinn, Menzie, and Jeffrey Frankel. 2008. “Why the Dollar Will Rival the Euro.” International Finance 11 (1): 4973.

  • Chiţu, Livia, Barry Eichengreen, and Arnaud Mehl. 2012. “When did the Dollar Overtake Sterling as the Leading International Currency? Evidence from the Bond Markets.” NBER Working Paper 18097, National Bureau of Economic Research, Cambridge, MA.

    • Search Google Scholar
    • Export Citation
  • Cohen, Benjamin J. 2015. Currency Power: Understanding Monetary Rivalry. Princeton, NJ: Princeton University Press.

  • Committee on Payments and Market Infrastructures (CPMI). 2020. Enhancing Cross-Border Payments: Building Blocks of a Global Roadmap—Stage 2 report to the G20, July.

    • Search Google Scholar
    • Export Citation
  • Dooley, Michael P., J. Saul Lizondo, and Donald J. Mathieson. 1989. “The Currency Composition of Foreign Exchange Reserves.” IMF Staff Papers 36 (2): 385434.

    • Search Google Scholar
    • Export Citation
  • Eichengreen, Barry, and Marc Flandreau. 2009. “The Rise and Fall of the Dollar, or When did the Dollar Replace Sterling as the Leading Reserve Currency?European Review of Economic History 13 (3): 377411.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Eichengreen, Barry, and Marc Flandreau. 2010. “The Federal Reserve, the Bank of England and the Rise of the Dollar as an International Currency, 1914–39.” BIS Working Papers 328, Bank for International Settlements, Basel.

    • Search Google Scholar
    • Export Citation
  • Eichengreen, Barry, and Donald J. Mathieson. 2000. “The Currency Composition of Foreign Exchange Reserves: Retrospect and Prospect.” IMF Working Paper 00/131, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Eichengreen, Barry, Arnaud Mehl, and Livia Chiţu. 2016. “Stability or Upheaval? The Currency Composition of International Reserves in the Long Run.” IMF Economic Review 64 (2): 35480.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Eichengreen, Barry, Arnaud Mehl, and Livia Chiţu. 2017. “Mars or Mercury? The Geopolitics of International Currency Choice.” NBER Working Paper 24145, National Bureau of Economic Research, Cambridge, MA.

    • Search Google Scholar
    • Export Citation
  • Eichengreen, Barry. 2019. Globalizing Capital: A History of the International Monetary System. Third Edition, Princeton, NJ: Princeton University Press.

    • Search Google Scholar
    • Export Citation
  • Erbenová, Michaela, Yan Liu, Nadim Kyriakos-Saad, Alejandro López-Mejía, Giancarlo, Gasha, Emmanuel Mathias, Mohamed Norat, Francisca Fernando, and Yasmin Almeida. 2016. “The Withdrawal of Correspondent Banking Relationships: A Case for Policy Action.” IMF Staff Discussion Note, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Financial Stability Board (FSB). 2020a. Addressing the Regulatory, Supervisory and Oversight Challenges Raised by Global Stablecoin Arrangements, FSB Consultative Document, Basel.

    • Search Google Scholar
    • Export Citation
  • Financial Stability Board (FSB). 2020b. “Enhancing Cross-Border Payments—Stage 1 Assessment Report to the G20.” Basel.

  • Financial Stability Board (FSB). Forthcoming. Enhancing Cross-Border Payments—Stage 3 Roadmap Report to the G20.” Basel.

  • Flandreau, Marc, and Clemens Jobst. 2009. “The Empirics of International Currencies: Network Externalities, History and Persistence.” The Economic Journal 119 (537): 64364.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Frankel, Jeffrey. 2012. “Internationalization of the RMB and Historical Precedents.” Journal of Economic Integration 27 (3): 32965.

  • Ghosh, Atish R., Jonathan David Ostry, and Charalambos G. Tsangarides. 2010. “Exchange Rate Regimes and the Stability of the International Monetary System.” IMF Occasional Paper 270, International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Goldberg, Linda, and Cedric Tille. 2013. “A Bargaining Theory of Trade Invoicing and Pricing.” NBER Working Paper 18985, National Bureau of Economic Research, Cambridge, MA.

    • Search Google Scholar
    • Export Citation
  • Gopinath, Gita. 2015. “The International Price System.” NBER Working Paper No. 21646, National Bureau of Economic Research, Cambridge, MA.

    • Search Google Scholar
    • Export Citation
  • Gopinath, Gita, Emine Boz, Camila Casas, Federico J. Díez, Pierre-Olivier Gourinchas, and Mikkel Plagborg-Møller. 2020. “Dominant Currency Paradigm.” American Economic Review 110 (3): 677719.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gopinath, Gita, and Jeremy C. Stein. 2018. “Trade Invoicing, Bank Funding, and Central Bank Reserve Holdings.” AEA Papers and Proceedings 108: 54246.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Heller, H. Robert, and Malcolm Knight. 1978. “Reserve Currency Preferences of Central Banks.” Essays in International Finance 131, Princeton University, Princeton, NJ.

    • Search Google Scholar
    • Export Citation
  • Horn, Sebastian, Carmen M. Reinhart, and Christoph Trebesch. 2019. “China’s Overseas Lending.” NBER Working Paper 26050, National Bureau of Economic Research, Cambridge, MA.

    • Search Google Scholar
    • Export Citation
  • Horsefield, Keith. 1969. The International Monetary Fund 1945–1965: Twenty Years of International Monetary Cooperation: Chronicle. Washington, DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Ilzetzki, Ethan, Carmen M. Reinhart, and Kenneth S. Rogof. 2019. “Exchange Arrangements Entering the Twenty-First Century: Which Anchor will Hold?The Quarterly Journal of Economics 134 (2): 599646.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (IMF). 2016. Strengthening the International Monetary System—A Stocktaking. IMF Policy Paper, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Ito, Hiro, and Robert N. McCauley. 2019. “A Key Currency View of Global Imbalances.” Journal of International Money and Finance 94 (C): 97115.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ito, Hiroyuki, Robert N. McCauley, and Tracy Chan. 2015. “Currency Composition of Reserves, Trade Invoicing and Currency Movements.” Emerging Markets Review 25 (C): 1629.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ito, Takatoshi, Satoshi Koibuchi, Kiyotaka Sato, and Junko Shimizu. 2019. “Growing Use of Local Currencies in Japanese Trade with Asian Countries: A New Puzzle of Invoicing Currency Choice.” Vox, June 10.

    • Search Google Scholar
    • Export Citation
  • Kindleberger, Charles. 1970. Power and Money: The Economics of International Politics and the Politics of International Economics. New York: Basic Books, Inc.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Laser, Falk Hendrik, and Jan Weidner. 2020. “Currency Compositions of International Reserves and the Euro Crisis.” Darmstadt Discussion Papers in Economics 238, Darmstadt University of Technology, Department of Law and Economics, Darmstadt.

    • Search Google Scholar
    • Export Citation
  • Li, David D., and Linlin Liu. 2008. “RMB Internationalization: Empirical and Policy Analysis.” In Currency Internationalization: Global Experiences and Implications for the Renminbi, edited by Wensheng Peng and Chang Shu, 16785. London: Palgrave Macmillan.

    • Search Google Scholar
    • Export Citation
  • Liao, Stephen, and Daniel McDowell. 2016. “No Reservations: International Order and Demand for the Renminbi as a Reserve Currency.” International Studies Quarterly 60 (2): 27293.

    • Search Google Scholar
    • Export Citation
  • Lindert, Peter H. 1969. “Key Currencies and Gold: 1900–1913.” Essays in International Finance 24, Princeton University, Princeton, NJ.

    • Search Google Scholar
    • Export Citation
  • Maddison Project Database. 2018. Groningen Growth and Development Centre, University of Groningen, Groningen, Netherlands.

  • Maggiori, Matteo, Brent Neiman, and Jesse Schreger. 2019. “The Rise of the Dollar and Fall of the Euro as International Currencies.” AEA Papers and Proceedings 109 (May): 521–26

    • Search Google Scholar
    • Export Citation
  • McKinsey & Co. 2016. Global Payments 2016: Strong Fundamentals Despite Uncertain Times.

  • Morahan, Aideen, and Christian Mulder. 2013. “Survey of Reserve Managers: Lessons from the Crisis.” IMF Working Paper 13/99, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Papamichalis, Theofanis, Camilo E. Tovar, Juan P. Treviño, Johannes Eugster, and Nathan Porter. Forthcoming. “A Multinetwork Model of Global Contagion.”

    • Search Google Scholar
    • Export Citation
  • Posen, Adam S. 2008. “Why the Euro will Not Rival the Dollar.” International Finance 11 (1): 75100.

  • Quinn, Stephen, and William Roberds. 2014. “Death of a Reserve Currency.” Federal Reserve Bank of Atlanta Working Paper 2014–17, Atlanta.

    • Search Google Scholar
    • Export Citation
  • Triffin, Robert, 1960. Gold and the Dollar Crisis: The Future of Convertibility. New Haven, CT: Yale University Press.

  • Voeten, Erik. 2013. “Data and Analyses of Voting in the United Nations General Assembly.” In Routledge Handbook of International Organization, edited by Bob Reinalda, 5466. London: Routledge.

    • Search Google Scholar
    • Export Citation
  • Voeten, Erik, Anton Strezhnev, and Michael Bailey. 2009. “United Nations General Assembly Voting Data.” https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/LEJUQZ.

    • Search Google Scholar
    • Export Citation
1

In this paper, the terms “country” and “economy” do not in all cases refer to a territorial entity that is a state as understood by international law and practice. These terms cover some territorial entities that are not states but for which statistical data are maintained on a separate and independent basis.

1

IMF Balance of Payments and International Investment Position Manual, sixth edition (BPM6).

2

”Reserve currencies” for the purpose of this paper are the currencies separately identified and reported in the IMF COFER database: eight currencies currently in use (the SDR currencies—US dollar, euro, Japanese yen, British pound, and Chinese renminbi, plus the Swiss franc, Canadian dollar, and Australian dollar—comprising 97 percent of total allocated reserves), and three currencies preceding and later replaced by the euro (the Deutsche mark, French franc, and Dutch guilder).

3

The Triffin dilemma refers to the fundamental tension between the heightened global demand for reserve currencies and the domestic policy incentives of reserve issuers, with implications for global financial stability. As such, the outsized role of the US dollar as a reserve currency was seen to impart instability in the system.

4

A 2012 IMF survey of reserve managers showed that many central banks were contemplating shifts to currencies such as the Australian and Canadian dollars (Morahan and Mulder 2013).

5

For instance, Ito and others (2019) show that the share of renminbi invoicing in Japanese exports to China increased from 1.3 percent in 2009 to 12.3 percent in 2017.

1

Many national authorities report the currency composition of their reserves to the IMF on a confidential basis reflecting market and/or political sensitivities.

2

Data from Eichengreen, Mehl, and Chiţu (2016) are originally sourced from IMF Annual Reports and Horsefield (1969).

3

The remaining 6 percent of global reserves are the unallocated reserves for which the currency breakdown is not available.

4

The reported shares of the US dollar and British pound cover the entire period of analysis (1947–2018). Other currencies cover shorter periods consistent with their status of “reserve currencies,” including the French franc and Deutsche mark since 1970 and Dutch guilder since 1973 (all three were replaced by the euro in 1999), Swiss franc and Japanese yen since 1973, Australian dollar and Canadian dollar since 2012, and Chinese renminbi since 2016.

5

The list of countries, year coverage, and sources are provided in Annex 3 Table 8. The sample consists of 15 countries in Europe, 8 in the Americas, 8 in Africa, 5 in Asia, and 6 in the Middle East and Central Asia. The panel is unbalanced; for example, only 8 countries report data for the full period and 3 for less than 10 years. In addition, the number of currencies reported varies by country, with some countries reporting separately only a few currencies. Limiting the sample to US dollar and euro shares, the two currencies consistently reported by most countries in the sample, yields qualitatively similar results.

6

Six out of 10 AEs in the sample are in Europe, whereas the comparable figure for EMDEs is 9 out of 32.

7

In contrast, Eichengreen, Mehl, and Chiţu (2016) rely on a random effects model.

8

An alternative measure capturing the financial links is the currency breakdown of external debt liabilities constructed by Bénétrix and others (2019). However, using this measure will further limit the sample (to 23 countries).

9

The geopolitical influence is proxied by several measures, including the proportion of countries that have voted in the same direction as the reserve issuer at the UN General Assembly in a given year; spending on official development assistance as a share of GDP; and military spending as a share of total military spending by reserve issuers (see Annexes 2 and 3 for further details).

10

The authors find that trade links have become more important for AEs since the GFC, driven by European countries in their sample. They also find that trade links are important for EMDEs with lower levels of total reserves (Annex 3).

1

As with past crises, the pandemic triggered a global market selloff and capital flight to safety. But capital outflows from EMDEs—at more than $100 billion in just two months—were more than three times larger than those seen during the GFC. At the same time, the massive capital outflows were short-lived and have already been partially reversed: exchange rates have stabilized, net issuance of bonds abroad reached $77 billion in April and May, and nonresident portfolio flows to EMDEs were back in positive territory in the second quarter of 2020.

2

In theory, data on individual countries’ reserves composition allow for a deeper investigation of triggers that make central banks drastically alter their holdings of one currency versus the other as reserves. However, such exploration is not feasible in this paper due to the short time span of the data and a limited number of such episodes.

3

The choice of euro for these thought experiments is dictated by data availability, which relates to its broad use as an international currency over the past 20 years. For opposite reasons, a similar exercise on the renminbi is not feasible at this stage. The magnitude of any increase is influenced by past trends: euro saw its share in cross-border bank claims increase by 15 percentage points within a 10-year period after its launch in 1999.

4

For instance, more localized production could reduce international trade and subsequently the demand for international reserves. Alternatively, more diversified international supply chains might encourage demand for a more diversified portfolio of reserves.

5

The COVID-19 crisis may prompt actions toward overcoming such impediments. For instance, the European Commission’s “next generation EU” proposal moots significant EU debt issuance over 2021–23.

6

The US Federal Reserve acted quickly to help support the smooth functioning of financial markets by activating bilateral swap lines with several major central banks, including in emerging markets, and creating an international repo facility for foreign monetary authorities.

7

For instance, the CFA franc was established in French African colonies after World War II; it was initially pegged against the French franc and subsequently against the euro. Countries using the CFA franc have been obliged to keep half of their reserves at the French treasury and to have a French representative on the currency union board in exchange for French guarantees on their balance of payments needs. Recent moves have been taken, however, to loosen such historical obligations—which may have implications for not only the location of reserves, but also their composition.

8

The internationalization of the renminbi has proceeded along two main lines. First, domestic and foreign companies have been encouraged to use renminbi in trade settlements, with the expectation that use in financial transactions will follow. This was followed, in 2018, by the launch of renminbi-denominated oil futures contracts, widening the scope for renminbi-denominated commodity trading. At the same time, there were efforts to boost the development of offshore renminbi markets and financial clientele and promote currency swap lines.

9

For instance, disagreements between the United States and European countries regarding the sanctions imposed on Iran prompted Germany, France, and the United Kingdom to create a parallel payment system (INSTEX) in 2019, which circumvents the dollar-dominated SWIFT messaging system and allows European companies to trade with Iran without risking to be sanctioned by the United States. INSTEX has concluded its first transaction in March 2020 by facilitating the export of medical goods from Europe to the pandemic-hit Iran.

10

Including 21 advanced and 45 emerging market economies.

11

For instance, Eastern Caribbean Central Bank has accelerated its plan to issue a CBDC by 2021.

12

PDCs can take various forms with differences in design and stability of value. For instance, while first generation crypto-assets (for example, Bitcoin and Ripple) are denominated in their own unit of account and exhibit large price volatility, stablecoins seek to minimize price fluctuations, enhancing their potential as a store of value.

13

The Financial Stability Board has developed a set of high-level principles for the regulation of GSCs, responding to a call by the G20 to examine regulatory issues and advise on multilateral responses, as appropriate (FSB 2020a).

14

Having multiple currencies could raise transaction costs to some extent. In the free banking period in mid-19th century, note reporters (periodicals) and brokers quoted secondary market prices for banknotes to help individuals identify and value various notes. Today, such information can be collated and shared almost instantaneously, aiding price discovery.

15

Large corporations and other strategic planners, as well as the IMF (Behar, Kostial, and Ramírez 2018), have increasingly used scenario planning to conceive plausible future states of the world.

1

Network effects could promote the use of a new reserve currency (by reducing the switching costs) once it reaches a critical mass or create a lock-in effect for an incumbent currency used widely because of high switching costs.

2

For example, the Federal Reserve system acted as a market maker for the US dollar. On the other hand, capital controls were used to limit access to the Deutsche mark in the 1960s and 1970s to better control inflation and allay exporters’ fears of loss of international competitiveness, while the internationalization of the Japanese yen also occurred despite initial domestic political resistance—the Foreign Exchange Law of 1980 allowed capital controls.

3

Chiţu, Eichengreen, and Mehl (2012) and Eichengreen and Flandreau (2010) also provide evidence that the development of US financial markets supported the increased role of the US dollar in trade finance and international debt markets.

4

Historical (political and economic) ties continued to support the sterling area and the international role of the British pound despite the declining role of the United Kingdom in the global economy. More specifically, after the United Kingdom left the gold standard in 1931, it encouraged key trading partners and colonies to peg their currencies against the pound to facilitate trade. Following World War II, the sterling area was formalized into a legally defined group with pegged exchange rates to sterling, common exchange controls against the rest of the world, and the maintenance of national reserves in sterling. Despite episodes of sterling devaluation, in 1970 the sterling area still comprised the United Kingdom and 35 other countries together with all British dominions, protectorates, protected states, and trust territories except Canada and Zimbabwe. The sterling area effectively dissolved with the demise of the Bretton Woods system in 1972.

5

Survey data also confirm that for EMDEs, the currency composition of reserves is driven by the currency composition of external liabilities, the composition of trade, and currency pegs (Morahan and Mulder 2013). For AEs, depth and liquidity of markets are the dominant considerations.

6

More recently, Laser and Weidner (2020) confirm the earlier findings, using the same methodology as Eichengreen and Mathieson (2000) but employing country-specific data on currency composition of reserves disclosed by various central banks. The methodology used in these papers is not robust to various specifications and does not account for the inertial effects.

7

Gopinath (2015) highlights that, in a sample of 43 economies, the dollar’s share for imported goods invoicing is about 4.7 times the share of US goods in imports.

8

The choice of the invoicing currency itself is influenced by the size and centrality of countries in global trade networks reflecting natural advantages, and the coalescence of exporters to limit competitive disadvantages (Bacchetta and van Wincoop 2005; Goldberg and Tille 2013).

1

The analysis presented here uses reserve shares unadjusted for valuation effects; however, the results are robust to using the valuation adjusted shares of Eichengreen, Mehl, and Chiţu (2017).

2

The main findings are robust to using, as a dependent variable, the share of synthetic euro reserves, which aggregates the shares of pre-1999 legacy currencies (Deutsche mark, French franc, Dutch guilder).

3

The high R-squared in all tables using aggregate reserve shares is due to the latter being a very slow-moving variable, which provides yet another reason to examine the reserve shares at country level.

4

The authors use a measure of the “eigenvector centrality” of the reserve issuer in the global trade network.

5

In using measures of financial development, the time horizon of the sample is substantially reduced due to their limited availability. Findings are robust to controlling for the GDP share alongside measures of financial development.

6

For more information on the UN voting data set, see “Data and Analyses of Voting in the UN General Assembly,” Voeten (2013).

7

A similar pattern emerges when average reserve shares are plotted against official development assistance; the US ODA as a share of GDP has fallen since the 1960s.

1

Time-series data for trade by invoicing currency are not available for many countries.

2

Other reserve currency shares, including renminbi shares, are not included due to the lack of data on cross-border bank claims and external public debt denominated in those currencies.

3

The currency denomination of public debt and cross-border bank claims are obtained from the World Bank International Debt Statistics dataset and the BIS international banking data by location, respectively.

4

In contrast from previous studies that use disaggregated data, the authors include the lagged reserve share as a regressor, which makes the panel dynamic and introduces dynamic panel bias, that is, strict exogeneity of the regressors no longer holds. The fixed effects model is no longer consistent when the number of country-currency pairs tends to infinity and T is fixed, while the interpretation of the random effects model depends on the assumption of initial values of a dynamic process. The Arellano-Bond estimator overcomes these issues but is designed for “small T large N” panels, which might not be applicable in this case given relatively small N. Also, all these models ignore the fractional nature of the dependent variable, that is, predicted values should always lie in the unit interval. A Tobit model addresses this issue but might suffer from the incidental parameters problem in the presence of fixed effects.

5

The authors do not find similar effect for financial links.

6

Using the de jure naturally restricts the sample to EMDEs. The de facto anchoring measure by Ilzetzki, Reinhart, and Rogoff (2019) based on exchange rate comovements has the drawback that it classifies some floating exchange rates as anchored to other currencies, for example, the Canadian dollar as anchored to the US dollar and the Swiss franc to the Euro.

Reserve Currencies in an Evolving International Monetary System
Author: Alina Iancu, Gareth Anderson, Mr. Sakai Ando, Ethan Boswell, Mr. Andrea Gamba, Shushanik Hakobyan, Ms. Lusine Lusinyan, Mr. Neil Meads, and Mr. Yiqun Wu