The Stealth Erosion of Dollar Dominance


Download The Stealth Erosion of Dollar Dominance


Preview text

The Stealth Erosion of Dollar Dominance:
Active Diversifiers and the Rise of Nontraditional Reserve Currencies
Serkan Arslanalp, Barry Eichengreen, and Chima Simpson-Bell WP/22/58
IMF Working Papers describe research in progress by the author(s) and are published to elicit comments and to encourage debate. The views expressed in IMF Working Papers are those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management.
2022
MAR

© 2022 International Monetary Fund

WP/22/58

IMF Working Paper Statistics Department

The Stealth Erosion of Dollar Dominance: Active Diversifiers and the Rise of Nontraditional Reserve Currencies

Prepared by Serkan Arslanalp, Barry Eichengreen, and Chima Simpson-Bell

Authorized for distribution by Carlos Sánchez-Muñoz March 2022

IMF Working Papers describe research in progress by the author(s) and are published to elicit comments and to encourage debate. The views expressed in IMF Working Papers are those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management.

ABSTRACT: We document a decline in the dollar share of international reserves since the turn of the century. This decline reflects active portfolio diversification by central bank reserve managers; it is not a byproduct of changes in exchange rates and interest rates, of reserve accumulation by a small handful of central banks with large and distinctive balance sheets, or of changes in coverage of surveys of reserve composition. Strikingly, the decline in the dollar’s share has not been accompanied by an increase in the shares of the pound sterling, yen and euro, other long-standing reserve currencies and units that, along with the dollar, have historically comprised the IMF’s Special Drawing Rights. Rather, the shift out of dollars has been in two directions: a quarter into the Chinese renminbi, and three quarters into the currencies of smaller countries that have played a more limited role as reserve currencies. A characterization of the evolution of the international reserve system in the last 20 years is thus as gradual movement away from the dollar, a recent if still modest rise in the role of the renminbi, and changes in market liquidity, relative returns and reserve management enhancing the attractions of nontraditional reserve currencies. These observations provide hints of how the international system m ay evolve going forward.
RECOMMENDED CITATION: Arslanalp, S., Eichengreen, B., and Simpson-Bell, C. 2022. The Stealth Eroson of Dollar Dominance: Active Diversifiers and the Rise of Nontraditional Reserve Currencies. IMF WP/22/58, Washington DC: USA

JEL Classification Numbers: Keywords: Author’s E-Mail Address:

F3, F31, F33
International reserves, currency composition, dollar
Serkan Arslanalp [email protected] Barry Eichengreen [email protected] Chima Simpson-Bell [email protected]

WORKING PAPERS
The Stealth Erosion of Dollar Dominance:
Active Diversifiers and the Rise of Nontraditional Reserve Currencies
Prepared by Serkan Arslanalp, Barry Eichengreen, and Chima Simpson-Bell

INTERNATIONAL MONETARY FUND

2

Contents
I. Introduction........................................................................................................................................................................ 5 II. Changes in the Dollar’s Share.....................................................................................................................................10 III. Analysis of Global Aggregates...................................................................................................................................14 IV. Analysis of Country-Level Data .................................................................................................................................17 V. Rise of Nontraditional Reserve Currencies...............................................................................................................22 VI. Robustness Checks.....................................................................................................................................................28
A. Large Reserve Holder Effects.............................................................................................................................28 B. COFER Reporting Effects....................................................................................................................................29 C. Exchange Rate Effects.........................................................................................................................................30 D. Interest Rate Effects.............................................................................................................................................32 E. Evidence of Rebalancing? ...................................................................................................................................33 VII. Conclusion...................................................................................................................................................................35 Ref erences ......................................................................................................................................................................... 37
FIGURES 1. Currency Composition of Global Foreign Exchange Reserves................................................................................ 6 2. Standard Determinants of US Dollar Share of Reserves........................................................................................10 3. US Dollar Index and Average Rate of Appreciation of Reserve Currencies vis-à-vis the SDR Basket, 1999– 2020.....................................................................................................................................................................................11 4. Additional Determinants of US Dollar Share of Reserve, 1999–2020...................................................................12 5. Switzerland and China: USD Share of Foreign Exchange Reserves....................................................................13 6. Emerging Market Economist: Excess International Reserves................................................................................13 7. Test for Structural Break in Issuer Size Coefficient..................................................................................................17 8. Determinants of US Dollar Shares..............................................................................................................................21 9. Official Reserve Shares of ‘Big Four’ Currencies vs. Nontraditional Currencies.................................................24 10. Bid-ask spreads of reserve currencies.....................................................................................................................26 11. Total Foreign Exchange Turnover between Nontraditional Currencies..............................................................27 12. Sharpe Ratios for Reserve Currencies....................................................................................................................28 13. US Dollar Share of Aggregate Official FX Reserves, Including and Excluding the Reserves of the Swiss National Bank, 1999–2020...............................................................................................................................................29 14. Allocated Reserves as a Share of Global Foreign Exchange Holdings..............................................................30 15. Exchange-Rate-Adjusted Dollar Share of Global Reserves, 1999-2020............................................................31 16. Maturity Composition of Foreign Official Sector Holdings of U.S. Treasury bonds, 2004–2020.....................32 17. Exchange-Rate-and-Interest-Rate-Adjusted Dollar Share of Global Reserves, 1999-2020............................33
TABLES 1. Global Aggregate Regressions, 1970–2020 .............................................................................................................16 2. Country-Specific Tobit Regressions, 1999–2020 .....................................................................................................19 3. Country-Specific Seemingly Unrelated Regression, 1999–2020 ...........................................................................20 4. Determinants of Currency Shares with Inertia but No Cross-Currency Effects, 2000–2020..............................22 5. Nontraditional Currencies in Global Foreign Exchange Reserves, end-2020......................................................23 6. Foreign Exchange Reserves in Nontraditional Currencies, end-2020 ..................................................................25

INTERNATIONAL MONETARY FUND

3

7. Rebalancing of USD Shares Using the US Dollar Share of Foreign Exchange Reserves, 1999–2020...........35

INTERNATIONAL MONETARY FUND

4

I. Introduction
The literature on dollar dominance (e.g., Boz et al., 2020) emphasizes the outsized role of the U.S. currency in global markets. The dollar’s share in global trade invoicing, international debt, and cross-border non-bank borrowing outstrips the share of the United States in trade, international bond issuance, and cross-border borrowing and lending. The currency’s dominance has been resilient in the face of a declining U.S. share of global GDP. Dollar dominance survived the collapse of Bretton Woods (Gourinchas, 2021), while its shares of international debt and non-bank borrowing rose still further following the global financial crisis (Eren and Malamud, 2021).
Influential contributions (e.g., Prasad, 2014) argue that the dollar has been the dominant international currency by default. The absence of alternatives has allowed it to dominate international funding markets, trade invoicing and settlement, and foreign exchange reserves. Other currencies suffer from an inadequate supply of investment-grade government securities for investors to hold as safe assets and central banks to accumulate as reserves (Eichengreen and Gros, 2020). Or their liquidity and availability is limited by regulation, including capital controls (Prasad and Ye, 2013; and Sullivan, 2020). They do not benefit from the large installed base of transactions denominated in dollars. They therefore lack the complementarities and synergies of different cross-border uses benefiting the dollar (Gopinath and Stein, 2021).
But with the rise of the euro and the renminbi, this narrative continues, the situation may be poised to change. Starting in 2012, with Mario Draghi’s “do whatever it takes” pledge, the European Central Bank asserted its readiness to act as liquidity provider of last resort to markets in euro-denominated assets in countries using the euro (European Commission, 2018). In 2020, with the creation of the €850 billion European Recovery Fund, there arose the prospect of a growing stock of safe and liquid AAA-rated government securities to be held as reserves by central banks (Hudecz, Cheng, Moshammer and Raabe, 2021).
China, meanwhile, embarked on a process of currency internationalization, aided by growing imports and exports, Belt and Road investments, a global network of renminbi currency swaps and official clearing banks, and addition of the renminbi to the Special Drawing Rights (SDR) basket (Subacchi, 2016; Greene, 2021). The capstone on this evolution, it is said (Jia, 2021), is now issuance by China of a central bank digital currency, the e-CNY.
In this paper, we focus on the currency composition of international reserves. On this dimension, the dollar has not become more dominant. It has not even maintained the dominance of prior years. Figure 1 shows the currency composition of foreign exchange reserves according to the IMF’s Currency Composition of Official Foreign Exchange Reserves (COFER) survey. According to this source, the share of reserves held in U.S. dollars by central banks dropped by 12 percentage points since the turn of the century, from 71 percent in 1999 to 59 percent in 2021.1

1 Reporting to COFER and the IMF Reserve Data Template (see Sections 5 and 6) are based on actual holdings of currencies, not exposure to a currency including the impact on positions of currency derivatives. This is because reserve assets, to qualify, must be readily available and be recorded on the central bank’s balance sheet. Accordingly, currencies to be received on the maturation of forward contracts are not counted as reserves; only the market value of the contract is counted. This is also because when reserve managers settle the forward contracts, they may not actually receive the currencies but may pay/receive just the market value of the contract without an exchange of notional values. We discuss some additional limitations of these data below.

INTERNATIONAL MONETARY FUND

5

This decline is not an inadvertent byproduct of changes in exchange rates, interest rate levels, or interest rate differentials. To the contrary, reserve managers have tended to rebalance their portfolios, restoring prior currency shares, to offset such changes. The decline is not the result of reserve accumulation by a small number of large reserve holders with a preference for non-dollar currencies. Nor is it a figment of changes in country or currency coverage of surveys of reserve composition. Rather, it reflects active portfolio diversification by central bank reserve m anagers.
Figure 1. Currency Composition of Global Foreign Exchange Reserves 1999–2021 (in percent)

80%

80%

70%

70%

60%

60%

50%

50%

40%

40%

30%

30%

20%

20%

10%

10%

0%

0%

1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021

U.S. dollar

Euro

Yen

Pound sterling

Other

Sources: IMF Currency Composition of Official Foreign Exchange Reserves (COFER). Note: The “other” category contains the Australian dollar, the Canadian dollar, the Chinese renminbi, the Swiss franc and other currencies not separately identified in the COFER survey. China became a COFER reporter between 2015 and 2018.
Figure 1 shows that this decline in the dollar’s share is not a shift toward the euro, the British pound sterling and Japanese yen—the other currencies that, historically, have played a significant international role and, along with the dollar, that have comprised the basket making up the IMF’s Special Drawing Rights.2 Though there was a rise in the share of reserves in euros after the turn of the century, this increase was not sustained. This is contrary to widespread expectations that the euro would come to play a more consequential international role and challenge the dollar’s dominant reserve currency status (see e.g., Chinn and Frankel, 2007, 2008).3
As a result, the decline in the dollar’s share has been m atched by a rise in the share of what we refer to as nontraditional reserve currencies, defined as currencies other than the US dollar, euro, Japanese yen and British pound sterling. As shown in Figure 1, the share of nontraditional reserve currencies rose from negligible levels at the turn of the century to roughly $1.2 trillion and 10 percent of total identified reserves in 2021. Using

2 The renminbi was added to the basket in 2016; we analyze the implications of this below. 3 Chinn and Frankel’s forecast was predicated on the assumption that the UK would join the Euro Area.

INTERNATIONAL MONETARY FUND

6

data from the IMF’s Special Data Dissemination Standard (SDDS) and national sources, we show that the shift away from the dollar has been a quarter a shift into the Chinese renminbi and three quarters a shift into other nontraditional reserve currencies. The renminbi has long been thought to pose a challenge to the dollar (see Subramanian 2011 for an early statement). However, not only is it starting out well behind the greenback, but the shift out of dollars, in fact, is not overwhelmingly a shift into the renminbi, as we show below. Rather, the shift out of dollars is more substantially a shift into the currencies of smaller economies that, historically, had less of the scale and liquidity needed to constitute an attractive form of international reserves.
This shift into nontraditional reserve currencies is substantial, and it is also broad based. We identify 46 “active diversifiers,” defined as countries with a share of official reserves in nontraditional currencies of at least 5 percent at the end of 2020.
Three factors contribute to the growing footprint of nontraditional reserve currencies. First is the growing liquidity of markets in those currencies. Historically, only a handful of countries have possessed deep and liquid markets in domestic-currency assets open to the rest of the world. Foreign exchange dealers, able to find counterparties only in that same handful of currencies, quoted and transacted in a limited number of bilateral exchange rates. Canales-Kriljenko (2004, p.7) describes how the principal currencies traded in the foreign exchange markets of developing countries at the turn of the century were the dollar, euro, pound sterling, and Japanese yen (currencies sometimes labelled the “Big Four”). He illustrates this by explaining that the cheapest way of purchasing Canadian dollars with Mexican pesos was by first purchasing U.S. dollars with Mexican pesos and then using the U.S. dollars to purchase Canadian dollars, reflecting the high liquidity and low transactions costs of m arkets for U.S. dollars.4
But as transactions costs have fallen with the advent of electronic trading platforms and now automated market-making (AMM) and automated liquidity management (ALM) technologies for foreign exchange transactions, the savings associated with transacting in U.S. dollars are less. Meanwhile, a growing number of countries have developed markets for trading currencies other than the Big Four.5 In addition, the expanding global network of central bank currency swap lines (Aizenman, Ito, and Pasricha, 2021) has enhanced the ability of central banks to access currencies other than the ones they hold as reserves, weakening these links across m arkets and functions.6
Second, central bank reserve managers have become more active in chasing returns. Central banks have accumulated substantial portfolios of financial assets. The larger the portfolio, the more scope there is for financial gains (and also, to be sure, losses) from active reserve management. When reserves exceed the level associated with reserve adequacy, reserve managers come to distinguish different reserve tranches: the minimum required for reserve adequacy (the “liquidity tranche”), which should be held in liquid, low-risk assets; and the rest (the “investment tranche”), which can be more actively managed, with returns in mind, and

4 Hence, it would be more expensive, in terms of transactions costs, for the Bank of Mexico, entering the market, to purchase and hold a nontraditional reserve currency such as the Canadian dollar, as opposed to the U.S. dollar. Intervening to support the peso would require the Bank of Mexico to first sell Canadian dollars for US dollars, the intervention currency.
5 To continue with the example, Canada reports transactions of its dollars against the Big Four but also the Mexican peso, Australian dollar, Swiss franc, Swedish krona and Hong Kong dollar (as well as a miscellaneous category of “other currencies”) in response to the most recent BIS Triennial Survey. See Bank of Canada (2019). Bank of Mexico reports onshore transactions in Canadian dollars and Swiss francs (Bank of Mexico 2019). Numbers for Mexico are relatively small, but the practice is more widespread elsewhere: Australia for example reports over-the-counter foreign exchange transactions in 22 currencies against the Australian dollar. Cheung, McCauley and Shu (2019) document how a growing number of emerging-market currencies have come to be traded in financial centers in addition to just New York, London and Tokyo in recent years.
6 This is apt to be true for some but not all swap lines, since in some cases the swap is not of one country's currency for the other country's currency but for U.S. dollars (Truman, 2021).

INTERNATIONAL MONETARY FUND

7

invested in less liquid assets (Hentov et al. 2019).7 The liquidity tranche is used to finance ongoing deficits, service and redeem debt, and intervene in the foreign exchange market, and as such tends to be held in the same relatively liquid assets denominated in the same currencies against which intervention is conducted. The investment tranche can be placed in nontraditional instruments and currencies. As the investment tranche grows relative to the liquidity tranche, one should expect to see more diversification in this direction.8
Third, and relatedly, as yields on bonds issued by the governments of the Big Four countries have fallen toward zero, central bank reserve managers may have intensified their search for higher yielding alternatives.9 As we show below, Sharpe Ratios (returns adjusted for volatility) have been more attractive for nontraditional currencies than the Big Four at a variety of points in the last decade.10
In sum, several factors have combined to prompt the shift from dollars to nontraditional reserve currencies in recent decades. The literature on reserve diversification (potential as well as actual) has tended to emphasize policy initiatives on the part of the official sector that fly under the banner of, inter alia, “renminbi internationalization” and fostering internationalization of the euro to boost Europe’s “strategic autonomy” (Reuters, 2020; Economist, 2021). Our analysis suggests that market forces and incentives matter at least as m uch as these policy m easures.
Our paper builds on several related literatures. Most obvious is the literature on dollar dominance. Gopinath and Stein (2021), already mentioned, emphasize mutually reinforcing synergies between use of the dollar in trade and bank-intermediated capital flows, while Farhi and Maggiori (2017) emphasize complementarities between dollar invoicing and the demand for dollar-denominated assets. Closely related is the literature on network effects and liquidity in foreign exchange markets. The literature on network effects (e.g., Matsuyama, Kiyotaki, and Matsui,1993; Rey, 2001) suggests that it pays to use and therefore for central banks to hold the same national currency used and held by others engaged in international transactions, since only that currency, or small handful of currencies, is widely priced and accepted, and since foreign exchange transactions are costly. Ogawa and Muto (2018) focus on liquidity and suggest that only large economies possess deep and liquid markets open to the rest of the world, thereby rendering their currencies attractive as international reserves.
Skeptics of this view (e.g., Eichengreen, Mehl, and Chitu, 2018; Eichengreen, 2019) question the importance of these complementarities between different international uses of a currency. They argue that as financial markets and relations develop, the case for a central bank to hold its reserves in the same currency that exporters invoice or banks borrow becomes weaker; they envisage movement toward a more multipolar

7 Of the central bank reserve managers surveyed in Castelli and Salman (2021), 68 percent report tranching or segmenting their reserves. There exist a number of definitions and associated measures of reserve adequacy, starting with the Greenspan-Guidotti rule of reserves equal to the current account deficit plus short-term debt maturing within the next year. Reserve managers may of
course vary the size of the liquidity tranche with market conditions (see Central Banking 2021).
8 Consistent with this, Aizenman, Cheung, and Qian (2020), discussed below, provide direct evidence that reserve managers diversify away from the Big Four currencies as their reserve portfolios grow large. In addition, to the extent that central banks are engaging in less foreign exchange intervention, they may be more comfortable about holding a portion of the liquidity tranche in nontraditional instruments and currencies. Indeed, Adler et al. (2021) find that foreign exchange interventions by emerging market economies have declined following the global financial crisis.
9 Consistent with this hypothesis, Castelli and Salman (2021) report survey responses from 30 central banks indicating that reserve managers see low and negative yields in fixed income markets as a concern, and that they have responded by moving in the direction of more diversification of reserves.
10 Low returns on government bonds of the Big Four countries may also explain the increase in gold reserves of central banks in the course of the last decade (Kitagawa 2021), although analysis of central bank demand for gold is properly the subject of another paper.

INTERNATIONAL MONETARY FUND

8

(multiple-currency) international monetary and reserve system. They point to the declining importance of network effects in an increasingly high-tech financial world and emphasize changes in market technology that reduce the cost and increase the ease of transacting in nontraditional international currencies.
Our paper is directly related to the empirical literature on the composition of foreign reserves. Eichengreen and Frankel (1996), Eichengreen (1998), and Chinn and Frankel (2007) have used published COFER aggregates to model the determinants of reserve currency shares and forecast their future evolution. Dooley, Lizondo, and Mathieson (1989) and Eichengreen and Mathieson (2000) used confidential country-level data underlying the COFER data base to explore the determinants of country-level reserve currency shares.11 Eichengreen and Mathieson found a striking degree of stability in the currency com position of reserves between the 1980s and 1990s and highlighted the importance of trade flows and debt denomination as determinants of currency shares.
As an alternative to the COFER data, Iancu et al. (2020) analyze data published by central banks. They assemble data on the reserves of 42 countries for the period 1999–2018. They conclude that financial links play an increasingly important role in the currency composition of reserves and that inertia in reserve proportions remains important. They find little evidence that trade shares significantly affect reserve composition, regardless of whether the former is measured as the share of trade with the reserve-currency country or the share invoiced in its currency. Similarly, Ito and McCauley (2020) use data published by 58 central banks. They find that dollar invoicing of exports is an important determinant of dollar reserve shares, and that countries hold larger dollar shares when their dom estic currency co-m oves with the dollar.
In the paper most closely related to our own, Aizenman, Cheung, and Qian (2020) assemble data for 58 countries from reports to the IMF on reserves in the form of nontraditional currencies (other than the dollar, euro, yen, and sterling).12 They find that countries that trade more with the US, Euro Area, Japan, and UK and that peg to their currencies hold a larger share of Big Four reserves; in addition, central banks diversify away from the dollar and other traditional reserve currencies as their reserve portfolios grow large. However, these authors do not distinguish changes in the shares of the renminbi and other nontraditional reserve currencies, as we do here.
Finally, there is a literature on rebalancing by central bank reserve managers. Dominguez, Hashimoto, and Ito (2012) analyze aggregate reserve accumulation; they distinguish between active reserve accumulation, resulting from purchases and sales of reserve assets, and passive accumulation, resulting from returns on reserve assets. In contrast to that paper, we use this decomposition to study changes in the currency composition of reserves rather than changes in total reserves. We also use total return indices to estimate valuation changes in reserve assets, whereas they use balance of payments estimates. Chinn, Ito, and McCauley (2021) use their country-level sample to test for rebalancing: they reject both no rebalancing and full rebalancing, where the extent of rebalancing increases with country and reserve-portfolio size. Also related is Truman and Wong (2006), who analyze published data on reserve composition for 23 central banks for the period 2000–04. They compare changes in reserve shares with and without exchange-rate valuation effects but do not consider rebalancing per se. Utilizing this same sample, Wong (2007) finds evidence of partial

11 Access to such data for research purposes is now more restricted, explaining why we do not utilize it here. 12 This data base provides a breakdown for the share of reserves not held in the “Big Four” reserve currencies, and, in some cases, breakdown within the Big Four. The IMF publishes these data under the heading “Data Template on International Reserves and Foreign Currency Liquidity” https://data.imf.org/?sk=2DFB3380-3603-4D2C-90BE-A04D8BBCE237. In addition, in Section IV, we draw on more detailed data disclosed by 80 economies on the currency composition of reserves utilizing the dataset compiled by Ito and McCauley (2020) as updated by Chinn, Ito, and McCauley (2021), the IMF’s Data Template and central bank annual reports.

INTERNATIONAL MONETARY FUND

9

Preparing to load PDF file. please wait...

0 of 0
100%
The Stealth Erosion of Dollar Dominance