THIRD RAMP SURVEY Central Bank Reserve Management Practices INSIGHTS INTO PUBLIC ASSET MANAGEMENT   | Central Bank Reserve Management Practices 1 © 2021 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy, completeness, or currency of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Cover photo Gerardo Pesantez / World Bank. Sample credit for a non-WB image © Ami Vitale / Panos Pictures. Used with the permission of Ami Vitale / Panos Pictures. Further permission required for reuse. Sample credit for a WB image © Curt Carnemark / World Bank. Further permission required for reuse. Report design Spaeth Hill Central Bank Reserve Management Practices INSIGHTS INTO PUBLIC ASSET MANAGEMENT1 Abstract: The World Bank Treasury’s Reserve Advisory and Management Partnership (RAMP) conducted its third survey on reserve management practices in 2021. One hundred and nineteen central banks, from different regions, income groups, and reserve levels, contributed to the survey, which included questions on investment policies, asset allocation, risk management, environ- mental, social, and governance (ESG) investing, and business continuity. The pandemic under- lined the importance of safety and liquidity for reserve portfolios. We find that central banks maintained their conservative investment approach, focusing on high-quality fixed-income assets denominated in US dollars and euros. At the same time, against a backdrop of ultra-low interest rates in major economies, we also observe that central banks continued, in their search for yield, to gradually diversify their reserves into more currencies and asset classes within fixed income. Survey results also indicate that central banks’ risk management practices show room for improvement, especially in institutions that have expanded into nontraditional asset classes, including those that invest in corporate credit. Meanwhile, reserve managers could further en- hance internal risk and reporting practices to strengthen oversight. ESG investing is still rarely adopted by central banks, and fewer than a quarter of respondents have included ESG objectives in their investment policy. Crucially, this is largely explained by the focus of reserve portfolios on high-quality fixed-income assets, among which ESG instruments and strategies are rarely encountered. We learn that, in order to maintain business continuity, central banks implement- ed home-based work in 2020, but technological drawbacks and cybersecurity concerns tended significantly to obstruct any ambition to attain fully remote reserve management operations. The paper carries the names of the authors and should be cited accordingly. The findings, interpreta- tions, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations or those of the Executive Directors of the World Bank or the governments they represent. Prepared by Mikaela Ballon Carneiro, Philip Dongsoo Hong, Daniela M. H. Klingebiel, Carmen Mileva 1  Herrero Montes, and Marco Antonio Ruiz Gil, under the auspices of the World Bank Treasury’s Reserve Ad- visory and Management Partnership. An initial draft of this report benefited from comments by Eric Bouye, Steen Byskov, Mark Choi , Heung Sik Choo, Attila Juhasz, Mike McMorrow, Shengting Pan, James Seward, Rodrigo Silveira Veiga Cabral, Daniel Vela Baron. Abstract  | Central Bank Reserve Management Practices 4 1. Introduction The World Bank Treasury's Reserve Advisory and Management Partnership (RAMP) launched its third survey on central banks' reserve management prac- tices in the first quarter of 2021. The previous two surveys have allowed central banks to benchmark their practices against those of peer institutions.2 Moreover, the data has helped RAMP improve its technical assistance for its members. The goal of this third survey is to continue expanding the understanding of reserve management policies and practices. The three surveys cover essential areas of reserve management and examine their evolution over time. We conducted the third RAMP survey at the height of the COVID-19 pandemic. Reserve managers faced several unprecedented shocks in 2020. While many gov- ernments restricted economic activity and movement to save lives and to combat the spread of the virus, economic activity ground to a halt worldwide. To boost their economies—severely hit by the pandemic in 2020—many central banks in advanced and developing economies introduced unprecedented monetary policy including near-zero or negative policy rates and large asset purchase programs to provide liquidity. Some countries faced extraordinary liquidity needs in foreign currency and had to use their reserves to meet the demand for foreign currency. Additionally, reserve management was more challenging because interest rates de- creased sharply—to historically depressed levels—in developed economies, where most portfolios are invested. Finally, central banks had to adjust their operations to confront an extraordinary environment where most staff had to work from home. This survey continues to expand the understanding of reserve management prac- tices globally. This year, the survey questionnaire covered not only questions on (i) governance; (ii) asset allocation; (iii) portfolio management; and (iv) risk man- agement, but also entailed additional questions on (v) environmental, social, and governance (ESG); (vi) pandemic business continuity; (vii) audit; and (viii) public disclosure. The survey contained 41 questions. Some of the questions gave a pre- defined set of potential responses, while others requested specific data. 2 The Inaugural RAMP survey was published in 2019. Ninety-nine central banks responded to our first survey, reflecting data from the first quarter of 2018. The second survey was published in 2020 containing informa- tion from 105 central banks with data from the third quarter of 2019. Introduction  | Central Bank Reserve Management Practices 5 A record-breaking number of central banks contributed to this survey—119.3 Institutions from different regions, income groups, and reserve levels responded.4 Table 1 depicts relevant metrics5 and shows that reserve levels and reserve adequa- cy metrics vary considerably by category.6 Although most participants provided extensive information, some did not answer all questions. Accordingly, this report identifies the number of respondents in the text, charts, or tables when presenting survey results. Survey results are presented in an aggregate format to preserve anonymity.7 We aggregate data across different categories, including geographic region, country income group, reserves size, reserve adequacy levels, and monetary policy and exchange rate regimes. To identify changes between surveys and assess differences in the respondents' responses across surveys, we only consider central banks that responded to both surveys.8 The report presents the patterns that then emerged. Moreover, the Appendix shows detailed results for the most relevant questions. The percentages reported in this report are based on the number of respondents for each question, where 3  100 percent refers to the total number of respondents replying to a given question, this being either equal to or below 119. The number of respondents to each question is reported in each chart. This report uses the World Bank’s customized income group categories based on GNI per capita calculated 4  using the World Bank Atlas Method. It separates countries into “low income” (GNI per capita of US$1,035 or less in 2020); “lower middle income” (US$1,036 to US$4,045); “upper middle income” (US$4,046 to US$12,535); and “high income” groups (US$12,536 or more) (World Bank Group 2020). For the purpose of this analysis, “lower- middle-income” and “low-income” countries have been grouped into the same category. The definitions of the categories are set out in such a way that an approximately similar number of institu- 5  tions is represented in each category. The adequacy of central banks’ levels of foreign exchange reserves can be measured in various ways, includ- 6  ing coverage of imports and short-term debt obligations. Unless otherwise specified, this report uses the term “adequacy” to denote a central bank’s possession of sufficient reserve assets to execute its mandate and achieve its objectives. Confidentiality facilitates central banks’ participation and candid and comprehensive responses, given the 7  sensitive nature of their operations. Similarly, we only analyze responses across surveys when the survey questions are comparable. 8  Introduction  | Central Bank Reserve Management Practices 6 Table 1. Survey Participants’ Reserve Levels and Adequacy Metrics NUMBER OF AVERAGE GDP AVERAGE AVERAGE AVERAGE CENTRAL PER CAPITA TOTAL TOTAL IMPORT BANKS (CURRENT RESERVES RESERVES COVERAGE US$) (US$ MILLION) TO GDP (MONTHS) Geographic region Americas & 23 9,831 41,376 0.3 5.8 Caribbean Europe & 41 27,769 89,297 0.2 4.2 Central Asia Middle East & Africa 32 4,554 18,782 0.4 5.3 South & East Asia 23 14,932 321,990 0.4 6.9 and Pacific Country income group High income 44 36,851 137,436 0.3 4.3 Upper middle 34 6,245 164,279 0.4 6.7 income Lower middle & 41 1,965 25,409 0.2 5.3 low income Size of assets under management (US$) Less than 3 billion 30 10,061 1,222 0.2 3.4 3 to 10 billion 31 9,498 6,297 0.2 4.9 10 to 50 billion 25 15,657 27,733 0.3 5.6 More than 50 33 26,384 362,272 0.4 7.0 billion Foreign exchange regime Not applicable 2 4,310 896 0.1 1.7 Floating 60 24,449 129,882 0.2 5.0 Soft Peg 50 5,474 87,802 0.4 6.0 Hard Peg 7 16,260 77,775 0.6 5.0 Grand Total 119 15,668 106,937 0.3 5.3 Source: IMF’s World Economic Outlook with data as of 2020, and World Bank’s World Development Indicators with data as of 2019. Notes: GDP = gross domestic product. a. Total reserves comprise holdings of monetary gold, special drawing rights, reserves of International Monetary Fund members held by the International Monetary Fund, and holdings of foreign exchange under the control of monetary authorities. The gold component of these reserves is valued at year-end (December 31) London prices. Data are in current US dollars. b. Respondents have been classified using the country income group categories defined by the World Bank as of June 2020. c. The respondent central banks’ assets under management (AUM) were categorized into four groups such that each group has a similar number of survey participants. d. All averages are simple averages (not weighted by reserves levels). Even though some central banks used reserves to provide foreign liquidity, at the global level, foreign reserves have continued to grow during the pandemic and Introduction  | Central Bank Reserve Management Practices 7 are now at historically high levels (Figure 1). Survey participants' reserve levels also grew on average by about 12 percent from the end of 2019 to the end of 2020. Figure 1. Accumulation of foreign reserves worldwide Volume of World Reserves 16 14 12 10 US$ trillions 8 6 4 2 0 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 Source: International Financial Statistics (IMF). Note: Last observation as of June 2021. Reserves relative to GDP - Largest emerging markets 60% International reserves (percent of GDP) 50% 40% 30% 20% 10% 0% Dominican Republic Brazil Kazakhstan Bulgaria Thailand Russia Algeria Peru Malaysia Morocco Iraq Vietnam Philippines Guatemala China Average EM India Columbia Romania Ukraine Mexico South Africa Bangladesh Turkey Indonesia Ghana Egypt Argentina Myanmar (Burma) Nigeria Kenya Ecuador Sri Lanka Pakistan Ethiopia Source: International Financial Statistics (IMF) as per World Bank country classification. Note: Data as of 2020. EM=emerging markets. This report has three parts. Section 2 presents the survey's key findings. Then, sec- tion 3 shows more detailed data and discusses significant patterns. Finally, section 4 explains the implications for policy engendered by these findings. Introduction  | Central Bank Reserve Management Practices 8 2. Key Findings The analysis of the survey reveals many critical findings and highlights some changes to reserve management practices. This section summarizes the main re- sults of our survey. Governance Concerning governance, self-insurance against external shocks remains the prima- ry objective for holding reserves. By contrast to the situation revealed by our 2018 survey, macroeconomic objectives have become more relevant—such as conducting foreign exchange policies and servicing international debt obligations. This could be related to the economic shock that most countries experienced in 2020. Most respondents identify safety and liquidity as the most critical principles in reserve management. Respondents deem these principles to be no less important than they were in 2018; meanwhile, the perceived relevance of income generation decreased. Despite the sharp reduction in interest rates in 2020, the economic crisis may have accentuated the importance of safety and liquidity relative to income generation. Investment policies: tranching and risk tolerance In their investment policies, three-quarters of respondent central banks divide their reserve portfolios into tranches in the strategic asset allocation (SAA) process, representing a 10 percent drop from our 2020 survey. Central banks with reserves above $50 billion and those located in Europe and Central Asia and high-income countries are less likely to use tranching. Differences in tranching practices may be the result of specific institutional and macroeconomic conditions. To define the institutional risk tolerance, Value at Risk (VaR), defined as the worst expected portfolio return, is the most common metric that central banks use. Al- most half of respondent central banks used this metric, while 43 percent of survey respondents utilize the probability of negative returns. More than 40 percent of central banks use more than one metric. It is also interesting to note that the preferred risk tolerance metric for high-in- come and upper-middle-income countries is VaR. In contrast, lower-middle and low-income countries prefer to use the probability of negative returns. By region, central banks in the Americas and the Caribbean prefer to use the probability of negative returns, while those in Europe and Central Asia prefer VaR. Key Findings  | Central Bank Reserve Management Practices 9 Strategic asset allocation Compared to our 2020 survey, we observe that central banks have reduced portfo- lio duration, most visibly in the investment tranche. This may indicate that central banks have sought to prepare their portfolios for a potential increase in interest rates once the economic impact of the pandemic subsides. In terms of currency composition, although the US dollar and the euro continue to be the most important currencies in reserve portfolios, central banks expanded eligibility and exposure to other currencies from 2020 to 2021. The most notable change in currency composition is a significant increase in the participation of the Chinese renminbi at the expense of the US dollar, the euro, and the British pound. The average allocation to the Chinese renminbi of respondent central banks amounted to two percent compared to an average 62.5 percent allocation to the US dollar and a 21.5 percent average allocation to the euro. In line with reserve management objectives, asset classes with high liquidity and low risk continue to be the most frequently used instruments in central bank port- folios. Nevertheless, most institutions have small allocations to nontraditional asset classes such as corporate bonds, emerging market bonds, covered bonds, mort- gage-backed securities, or equities.9 Individual allocations to these asset classes did not exceed five percent on average, compared to a 34 percent average allocation to bonds, a 23 average allocation to bank deposits, a 12 percent average allocation to supranational bonds, and a 10 percent average allocation to money market instru- ments. Respondents reported an average allocation to gold of eight percent. It is interesting to note that while 60 percent of central banks hold gold in their reserve portfolios, only 20 percent of them include it in their strategic asset allocation. In several countries, historical precedents or the requirement to buy gold from local producers may determine the level of gold holdings. By income level, the central banks of high-income countries have concentrated their portfolios in government bonds, while lower-middle- and low-income coun- tries tend to favor bank deposits. At the same time, the central banks of high-in- come countries have the highest average allocation to nontraditional asset class- es—13.8 percent on average—followed by their upper-middle-income counterparts (6.5 percent, on average) and lower-middle-income and low-income countries (3.6 percent). Finally, compared to our 2020 survey, central banks expanded eligibility and expo- sure to most asset classes within fixed income, indicating an unabated appetite for gradual diversification. The criterion for classifying asset classes as nontraditional was whether fewer than 50 percent of the re- 9  spondents reported the asset class as eligible. Key Findings  | Central Bank Reserve Management Practices 10 Use of external managers Central bank reserve managers continue to use external managers in their invest- ment operations. Three-quarters of respondents use external managers. Knowledge transfer and return enhancement are the main motivations to hire external man- agers. In addition, a significant number of central banks use external managers to create exposure to nontraditional asset classes. Between 2018 and 2021, the number of central banks with external management programs increased marginally from 69 to 72 percent. Management of credit and market risk Central banks continued to manage credit risk by investing mainly in debt with high credit ratings. As in previous surveys, most participants continue to have minimum ratings in the investment-grade category (BBB+/BBB/BBB- or above). In that category, the most frequent minimum credit rating for all asset classes is A+/A/A-. Rating agencies remain the primary source of information on credit risk, with almost all the survey respondents using credit ratings. In addition, 82 percent of respondents use other methodologies to assess credit risk. At the same time, 53 percent of central banks are not using aggregate credit risk measures. Without these measures, an institution is unable to assess whether credit risk in the portfo- lio is consistent with its institutional risk tolerance. Most central banks use duration limits to manage market risk, as reserve portfo- lios mainly consist of fixed income instruments. However, it is notable that many central banks do not use probabilistic risk metrics as hard limits, for example VaR and expected shortfalls (Conditional VaR). In addition, only 20 percent of central banks that deploy active risk use tracking error and a risk budgeting framework to monitor and manage their active decisions and ensure that they optimally deploy active risk. The detailed findings on risk management highlight that there is still room for cen- tral banks to strengthen their risk management framework. Enhancing risk man- agement practices is particularly important for central banks with corporate credit exposure, diversified portfolios in terms of currency and asset class, and active risk, including deviations from their strategic asset and currency benchmarks. Reporting on investment results to the board, invest- ment committee, and the public Survey results indicate that the investment committee usually receives more infor- mation on risk and performance than the governor or the board, but reports often lack some key metrics. The investment committee usually has more oversight over the day-to-day management of the portfolios, and it is reasonable that they receive more comprehensive reports. More than half of central banks include the most rel- Key Findings  | Central Bank Reserve Management Practices 11 evant risk and performance metrics in the investment committee reports. However, most central banks do not report tracking error, credit VaR, or rating breakdowns to the investment committee, the governor, or the board. Improving risk and per- formance reports, while avoiding excessive frequency or detail, is an essential step to enhance the governance of the investment process. Environmental, social, and governance (ESG) investing Survey answers to the questions on ESG reveal that the adoption of ESG investing in reserve management is still low, with fewer than a quarter of respondents hav- ing included ESG objectives in their investment policy. A critical limitation here is the focus of reserve portfolios on high-quality fixed-income assets, among which ESG instruments and strategies are rarely encountered. Having a positive impact and contributing to their reputation are the main drivers of ESG implementation in central banks. Impact investment through green, social, and sustainability bonds is a preferred ESG strategy for central banks that imple- ment this practice. Nonetheless, the proportion of portfolios invested in green, so- cial, and sustainability bonds is minimal, as can be expected given the very modest (albeit increasing) size of this market segment. Business continuity management The COVID-19 pandemic placed an enormous operational burden on most insti- tutions. From an operational perspective, roughly half the central banks were not prepared for a pandemic in 2019. Although most institutions changed their work arrangements in response to the pandemic, moving toward a fully remote opera- tion proved to be complex. Remote work arrangements from home were challeng- ing because certain critical functions, especially settlement and payment areas, could not be performed remotely. Cybersecurity and lack of appropriate hardware and software were the main constraints in that respect. Public disclosure Although most institutions produce internal reports, disclosing extensive infor- mation on reserve management activities to the public is not typical. The only data that most central banks (60 percent) reveal are the currency composition of reserves. A significant proportion also publishes eligible asset classes (50 percent), performance (44 percent), and asset allocations (41 percent). Fewer than a quarter of central banks divulge their investment policies, guidelines, risk metrics, or the characteristics of external management programs. Key Findings  | Central Bank Reserve Management Practices 12  esults and 3. R Observations This section provides the survey results and identifies prominent patterns. It also highlights notable changes compared to the previous survey results. The subsec- tions are: (i) governance; (ii) strategic and current asset and currency allocation; (iii) portfolio management; (iv) risk management; (v) ESG; (vi) pandemic business conti- nuity; (vii) audit, and (viii) public disclosure. 3.1. GOVERNANCE a. Investment objectives Central banks have various objectives for holding foreign exchange reserves. These include self-insurance against external shocks, conducting foreign exchange policy, servicing external debt or other obligations, and supporting monetary policy opera- tions. These objectives determine crucial decisions for reserve management, includ- ing risk tolerance, investment horizon, and currency and asset allocation. This survey found that self-insurance against external shocks remains the primary objective for holding reserves. Most respondents (80 percent) consider self-insur- ance a highly relevant objective (see Figure 2). Conducting foreign exchange policy (67 percent) and servicing external debt or other obligations (55 percent) are also highly relevant. By contrast, few central banks (seven percent) regard saving for intergenerational equity as an investment objective. Figure 2. Investment Objectives Highly relevant Somewhat relevant Not relevant Provide self-insurance against 80% 12% 8% potential external shocks Conduct foreign 25% 8% 67% exchange policy Service external debt 19% 23% 55% or obligations Support monetary 40% 24% 29% policy operations Ensure savings for 7% 22% 59% intergenerational equity Other 9% 4% 17% 0% 50% 100% 0% 50% 100% 0% 50% 100% Percentage of respondents N=119. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: The responses may not add up to 100 percent because some respondents to the question did not classify all choices presented. Results and Observations  | Central Bank Reserve Management Practices 13 Compared to our 2018 survey, the importance of pursuing macroeconomic ob- jectives through reserve accumulation increased while the intergenerational savings objective became less important for respondents.10 A higher proportion of institutions rank conducting foreign exchange policy, servicing external debt or obligations, and supporting monetary policy operations as highly relevant (Figure 3). On the other hand, the percentage of respondents indicating that saving for intergenerational equity is not relevant increased by seven percent. The COVID-19 pandemic was a black swan event that highlighted to central banks the importance of holding reserves to navigate a challenging macroeconomic environment. Figure 3. Changes in investment objectives Highly 2018 85% Provide self- relevant 2021 85% insurance Somewhat 2018 11% against relevant 2021 9% potential ext- ernal shocks Not 2018 0% relevant 2021 4% Highly 2018 67% Conduct relevant 2021 72% foreign Somewhat 2018 21% exchange relevant 2021 22% policy Not 2018 11% relevant 2021 6% Highly 2018 52% Service relevant 2021 56% external Somewhat 2018 24% debt or relevant 2021 22% obligations Not 2018 17% relevant 2021 18% Highly 2018 28% Support relevant 2021 30% monetary Somewhat 2018 48% policy relevant 2021 46% operations 2018 17% Not relevant 2021 16% Highly 2018 9% Ensure relevant 2021 6% savings Somewhat 2018 28% for interge- relevant 2021 24% nerational equity Not 2018 52% relevant 2021 57% Highly 2018 4% relevant 2021 6% Other Somewhat 2018 4% relevant 2021 2% Not 2018 5% relevant 2021 16% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of respondents common to both surveys N= 82. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: The responses may not add up to 100 percent because some respondents to the question did not classify all choices presented. This question was included in the first RAMP survey (2019), but not in the second (2020). 10  Results and Observations  | Central Bank Reserve Management Practices 14 b. Investment principles Safety and liquidity continue to be the essential principles of reserve manage- ment (Figure 4). More than 94 percent of respondents identified those principles as highly relevant. By contrast, income generation (returns) was highly relevant for only a third of central banks. For the majority (65 percent), income generation was somewhat relevant. As foreign reserves are a significant component of central banks' balance sheets, returns matter for these institutions, but less than safety and liquidity. Figure 4. Investment Principles Highly relevant Somewhat relevant Not relevant Safety 97% 3% 0% (preservation of capital) Liquidity 94% 5% 1% Returns 34% 65% 0% (income generation) Other 3% 3% 12% 0% 50% 100% 0% 50% 100% 0% 50% 100% Percentage of respondents N=119. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: The responses may not add up to 100 percent because some respondents to the question did not classify all choices. Respondents judged that safety and liquidity were as crucial in 2021 as in 2018, but the perceived relevance of income generation as an objective of reserve management activity decreased.11 More than 96 percent of central banks classified safety and liquidity as highly relevant in 2018 and 2021. However, the percentage of respondents that considered returns as highly relevant decreased slightly (by seven percent). Similarly, more respondents now consider income generation as some- what relevant (an eight percent increase).12 Despite the sharp reduction in interest rates in 2020 in major economies, the economic crisis and market volatility may have accentuated the importance of safety and liquidity for central banks in line with their objective, while income generation may now appear less significant. This question was included in the first RAMP survey (2019), but not in the second (2020). 11  We cannot discern any geographic pattern. 12  Results and Observations  | Central Bank Reserve Management Practices 15 3.2. STRATEGIC ASSET ALLOCATION a. Tranching Dividing reserve portfolios into tranches is common in the strategic asset alloca- tion process. Typically, central banks divide reserve portfolios into tranches—such as those for working capital, liquidity, and investment—which differ by investment objectives, liquidity needs, and investment horizon. Seventy-three percent of the respondents use this practice. Interestingly, the proportion of respondents that use tranching decreased by 10 percent from 2020 to 2021, explained mostly by more central banks responding to this question. As noted in the previous RAMP surveys, some central banks may find tranching helpful in enhancing diversification and lengthening the portfolio's investment horizon (the investment tranche). A tranch- ing framework may make it easier for staff to communicate objectives linked to specific portfolio tranches more easily to their board and investment committee. From a methodological standpoint, institutions can achieve both—diversifying their reserves and extending the investment horizon—without tranching. While most central banks use a tranching framework, we found substantial differ- ences across central banks. Figure 5 shows tranching by size, region, and income group. Most institutions with reserves between $3 and $10 billion, those located in the Americas and the Caribbean, Middle East and Africa, and those in lower-mid- dle and low-income countries used tranching. By contrast, central banks less likely to have significant short-term liquidity needs (reserves above $50 billion, located in Europe, or other high-income countries, and with floating exchange rate regimes) were less likely to tranche their reserve portfolios, which may explain why their portfolios carry more risk.13 Central banks without tranches have a longer average investment horizon (36 versus 25 months in central 13  banks with tranches), a higher average duration (23 versus 17 months), and a higher average allocation to nontraditional asset classes (11 percent versus 7 percent). Results and Observations  | Central Bank Reserve Management Practices 16 Figure 5. Tranching by size, region, and income group Tranching by income group Tranching by region Tranching by size 100% 100% 100% 13% 13% 9% 10% 10% 10% 90% 21% 21% 90% 17% 18% 90% 23% 23% Percent of country-income group 28% 28% 28% 28% 80% 80% 80% 45% 49% 42% 41% Percent of region group 45% 48% Percent of size group 70% 70% 70% 60% 60% 60% 50% 50% 50% 90% 87% 90% 40% 40% 82% 91% 40% 90% 88% 79% 83% 77% 77% 72% 72% 79% 72% 71% 30% 30% 30% 55% 59% 58% 55% 53% 20% 20% 51% 20% 10% 10% 10% 0% 0% 0% Europe & Lower middle and low income Upper middle income High income Americas & Caribbean Central Asia Middle East & Africa South & East Asia & Pacific Less than 3 billion 3 to 10 billion 10 to 50 billion More than 50 billion NO YES N=118. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. The survey results indicate that the investment tranche accounts for the largest share of reserves, followed by the liquidity and working capital tranches. Figure 6 shows a box plot with the distribution of the allocation of the tranches in reserve portfolios, identifying the minimum, the second quartile group, the median, the third quartile group, and the maximum of our survey respondents. The investment tranche accounted on average for 46 percent of total reserves. In contrast, the liquidity and working capital tranches accounted for an average of 39 and 19 per- cent, respectively. Empirical evidence indicates that central banks may determine the size of the liquidity tranche on the basis of reserve adequacy metrics.14 If that were to hold, then the considerable size of the average investment tranche would suggest that many respondent central banks have adequate reserves to cover short- and medium-term liquidity to manage their macroeconomic objectives. According to Heller 1966, there are three motives for holding international reserves: transactional, 14  speculative, or precautionary. Empirical results (see, for example, Aizenman 2007) support precautionary (versus mercantilist) motives. In this case, reserves play the role of self-insurance against sudden stops in capital flows. Jeanne and Rancière (2011) develop a model of the optimal level of international reserves for a small open economy. The model focuses on the benefits of holding reserves for crisis mitigation and the reduction of the probability of sudden stops and is consistent with the average level of reserves in emerging market countries since 1980, with the exception of Asia since 1998. Results and Observations  | Central Bank Reserve Management Practices 17 Figure 6. Participation of tranches in reserve portfolios 100% 90% Allocation by tranche (percent) 80% 70% 60% 50% 40% 30% 20% 10% 0% Working Liquidity Investment Other capital tranche tranche tranche Min 1.0% 3.0% 1.7% 2.1% Max 89.7% 99.6% 94.9% 51.0% Median 11.0% 34.5% 47.0% 11.0% Average 18.7% 39.4% 46.5% 14.8% Number of 49 77 73 23 respondents N=78. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: The distribution for each tranche is based on the total number responding to the question. The data are divided into four equal-sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The median marks the midpoint of the data and is displayed as the line between the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are smaller than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (lowest point). The horizontal red line refers to the average of each category. b. Risk Tolerance and Investment Horizon Almost half of the respondents use a Value at Risk (VaR) metric to explain the risk tolerance of their reserve management operations. VaR is defined as the worst expected return of the portfolio over a predefined probability and time horizon. Forty-three percent of institutions used the probability of negative returns, and 39 percent of respondents deployed Conditional Value at Risk (CVaR) (Figure 7) to state the risk tolerance of their reserve management operations. Roughly 40 percent of respondents use more than one metric.15 The preferred risk tolerance metric for high-income and upper-middle-income countries was VaR, while low- er-middle and low-income countries preferred to use the probability of negative 15  Setting a probability of negative returns at 5 percent is equivalent to a VaR of zero percent with 95 percent confidence. Results and Observations  | Central Bank Reserve Management Practices 18 returns. By region, central banks in the Middle East and Africa preferred to use the probability of negative returns. In contrast, those in Europe preferred VaR, which may reflect the yield level available in their major reserve currency. Finally, central banks with less than $10 billion in reserves prefer to use the probability of negative returns, while those with more than that amount prefer VaR. Figure 7. Metric to define risk tolerance in Strategic Asset Allocation (SAA) framework 60% 52% 50% 43% Percentage of respondents 39% 40% 30% 27% 20% 10% 0% Value at Conditional Probability Other risk (VaR) VaR (CVaR) of negative returns N=114. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Our survey finds that some central banks decided to increase their risk toler- ance levels during 2020, probably in response to the low-yield environment and the increased levels of market volatility. When asked whether central banks changed their risk tolerance measures in 2020, 15 percent of respondents that use the probability of negative returns reported increasing the threshold for negative returns. Fourteen percent of respondents increased the absolute level of VaR, and 10 percent increased the absolute level of CVaR (Figure 8). The small proportion of central banks that increased their institutional risk tolerance probably had to respond to the low-yield environment and the increased levels of market volatility to maintain their strategic asset allocation. Again, with yields closer to zero per- cent—or even below, in many major economies—there is an increased probability of negative returns on fixed-income instruments. Thus, the accrued interest is not enough to compensate for potential decreases in market prices. Events in the US Treasury markets in March 2020 and the bouts of market volatility during 2020 may explain why some institutions also decided to increase the threshold for their VaR and CVaR measures. When asked whether institutions added or excluded risk tolerance metrics, almost all institutions responded that they did not change the number of risk measures used. During 2020, nearly all institutions neither added new risk measures nor excluded existing ones. Results and Observations  | Central Bank Reserve Management Practices 19 Figure 8. Changes in thresholds of risk tolerance metrics Yes, it increased Yes, it decreased No Probability of 15% 7% 78% negative returns Value at risk (VaR) 14% 0% 86% Conditional VaR 10% 0% 90% (CVaR) Other 8% 0% 92% 0% 50% 100% 0% 50% 100% 0% 50% 100% Percentage of respondents N=115. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Not surprisingly, for central banks that use a tranching framework, the duration of the investment tranche is substantially higher than that of the liquidity port- folio. Reserve managers assign a different investment objective to the investment tranche. They typically use the tranche to construct investment portfolios with a longer investment horizon than the liquidity tranche to pursue higher returns.16 Our survey results confirm this notion. Survey respondents reported an average investment horizon of 42 months for their investment tranche compared to 13 months for the liquidity tranche (Figure 9). Siegel (2008) argues that a longer investment horizon goes hand in hand with an increasing ability to take 16  risk and increase expected returns. Results and Observations  | Central Bank Reserve Management Practices 20 Figure 9. Investment horizon of tranches and total portfolio 61% 60% percent Liquidity 40% tranche 30% 20% 4% 3% 1% 0% 60% 44% Investment percent 40% tranche 20% 20% 17% 14% 5% 0% 2% 60% 41% percent Total 40% 36% portfolio 20% 9% 9% 5% 0% 60% percent Total portfolio 40% 36% (untranched) 28% 20% 16% 12% 8% 0% Less than 12 Between 36 Between More than 12 months months 12 and 36 months 36 and 60 60 months months months N=111. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. In line with their portfolio objectives, central banks continue to hold portfolios of low duration, and the average duration of their total portfolio stood at 22 months at the end of 2020. Figure 10 illustrates that central banks without a tranching framework had a higher overall average portfolio duration (23 months) than those with tranching frameworks (17 months). Results and Observations  | Central Bank Reserve Management Practices 21 Figure 10. Duration of total portfolio and by tranche 120 110 100 90 80 Duration (months) 70 60 50 40 30 20 10 0 Liquidity Investment Total portfolio Total portfolio tranche tranche (untranched) Min 1.0 1.0 3.0 3.0 Max 60.0 120.0 48.0 72.0 Median 6.0 22.0 14.0 20.0 Average 8.4 25.9 17.1 22.6 Number of 69 69 29 31 respondents N=80 and N=31 for untranched. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: The distribution of the duration for each tranche is based on the total number responding to the ques- tion. This figure displays the minimum, the second quartile group, the median, the third quartile group, and the maximum of the duration to each tranche and the portfolio as a whole. The data are divided into four equal-sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The me- dian marks the midpoint of the data and is displayed as the line between the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are smaller than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (lowest point). The horizontal red line refers to the average of each category. Comparing survey results of 2021 with those of 2020, we find that central banks reduced portfolio duration significantly—by an average of seven months. Central banks without a tranching framework decreased the average duration of their port- folio slightly more, by eight months, on average (Figure 11). Central banks with a tranching framework mainly reduced the duration of the investment tranche while leaving that of the liquidity tranche essentially unchanged. The shortening of duration from 2020 to 2021 is probably a reaction to the ul- tra-low interest rates seen during most of 2020 and may therefore be temporary. As major central banks continued to use all of the tools at their disposal to provide monetary policy stimulus with the onset of COVID, yields on fixed-income securi- ties decreased substantially. By shortening the duration, reserve managers may be Results and Observations  | Central Bank Reserve Management Practices 22 preparing their portfolio for a potential increase in rates when major central banks unwind their unprecedented policies as the global economy recovers from the pandemic. Figure 11. Changes in duration (compared to the previous survey) 20 10 0 -10 Change in duration (months) -20 -30 -40 -50 -60 -70 -80 -90 -100 -110 -120 Liquidity Investment Total portfolio Total portfolio tranche tranche (untranched) Min -53.0 -120.0 -60.0 -81.0 Max 21.0 19.0 12.0 13.0 Median 0.0 0.0 0.0 0.0 Average 0.4 -7.2 -4.6 -8.4 Number of 48 48 21 15 respondents N=72. Source: Third RAMP survey on the Reserve Management Practices of Central Banks, Second RAMP survey on the Reserve Management Practices of Central Banks. Note: The distribution of the duration for each tranche is based on the total number responding to the ques- tion. This figure displays the minimum, the second quartile group, the median, the third quartile group, and the maximum of the duration to each tranche and the portfolio as a whole. The data are divided into four equal-sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The me- dian marks the midpoint of the data and is displayed as the line between the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are smaller than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (lowest point). The horizontal red line refers to the average of each category. c. Eligible Currencies and Actual Currency Composition of Reserves Some authors argue that the composition of the balance of payments and port- folio management objectives help explain the currency composition of reserve portfolios (Wang 2019). Here, the structure and denomination of external debt, Results and Observations  | Central Bank Reserve Management Practices 23 intervention needs, and asset and liability management may significantly influence the currency composition of reserves. On the other hand, central banks also con- sider the liquidity profile of assets denominated in different currencies to deter- mine eligible currencies (depending on their desire to diversify currency exposure). While the structure of the balance of payments is more important for the currency composition of the liquidity tranche, portfolio management concerns play a more critical role in shaping the design of the investment tranche (Alekasir et al. 2019). The US dollar continues to be the most widely allowed currency in reserve portfo- lios. Nearly all respondents indicated that they could invest in the US dollar. Most institutions also allow investments in the other currencies that make up the special drawing rights (SDR) basket: 88 percent can invest in the euro, 81 percent in the British pound, 65 percent in the Japanese yen, and 65 percent in the Chinese ren- minbi (Figure 12). Notably, the Australian and Canadian currencies are also eligible for more than half of central banks. Figure 12. Percentage of respondents that may hold each currency as part of their foreign exchange reserves USD 99% EUR 88% GBP 81% JPY 65% CNY 65% AUD 60% CAD 58% CHF 45% NOK 34% SEK 33% NZD 30% DKK 28% SGD 23% KRW 21% HKD 14% ZAR 11% MXN 10% INR 7% TRY 5% RUB 4% BRL 4% Other 32% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of respondents N=113. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Since our last survey in 2020, central banks have expanded the list of eligible currencies considerably. The Chinese renminbi (CNY) notably benefitted from this development. Now two-thirds of institutions allow CNY investments, on a par with the Japanese yen (JPY). The eligibility of the Chinese currency had increased by 15 percent compared to our last survey. At the same time, central banks also Results and Observations  | Central Bank Reserve Management Practices 24 expanded into other currencies and away from the US dollar and the euro (see Figure 13). Some notable examples of this trend include the Australian dollar, the Canadian dollar, and some Asian currencies, such as the Singaporean dollar or the Korean won. Figure 13. Changes in currency eligibility (compared to the previous survey) CNY 15% KRW 9% SGD 8% AUD 6% NZD 5% NOK 4% CAD 4% GBP % DKK 2% CHF 2% JPY 2% SEK 1% USD 0% -2% EUR -5% 0% 5% 10% 15% 20% Net change in percentage of respondents N=97. Source: Third RAMP survey on the Reserve Management Practices of Central Banks, Second RAMP survey on the Reserve Management Practices of Central Banks. The US dollar still predominated in reserve portfolios, followed by the euro. In 2021, respondent central banks allocated an average of 62.5 percent of their re- serves to the US dollar, followed by the euro, to which they assigned an average of 21.5 percent of reserves. These results are consistent with the IMF’s Currency Composition of Official Foreign Exchange Reserves (COFER),17 although it is worth noting that the IMF reports the aggregate currency composition across all countries while this survey analyzes the distribution of individual responses. Figure 14 shows the distribution range of the currency composition of all respondents' foreign ex- change holdings, including those that reported no allocation to the respective cur- rency. In the case of the euro, although a small number of central banks had large allocations to the currency, most respondents had small exposures or no allocation.18 The individual average allocations to all other currencies are still relatively low (be- low three percent). However, taken together, exposure to currencies other than the US dollar and the euro accounted for an average of 16 percent of reserve portfolios, suggesting that most portfolios have small exposures to multiple currencies. https://data.imf.org/?sk=E6A5F467-C14B-4AA8-9F6D-5A09EC4E62A4 17  The median allocation to the US dollar is 72 percent, and the average is 63 percent. In the case of the euro, 18  the median allocation is 5 percent, and the average is 22 percent. Results and Observations  | Central Bank Reserve Management Practices 25 Figure 14. Distribution of allocations to individual currencies 100 90 Actual currency allocation (percent) 80 70 60 50 40 30 20 10 0 USD EUR GBP CNY AUD JPY CAD SGD KRW NOK NZD ZAR DKK INR SEK HKD RUB BRL CHF MXN TRY Other Min 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Max 100.0 100.0 23.0 20.0 32.0 18.0 13.0 3.0 5.0 6.0 23.0 74.0 1.0 25.0 1.0 1.0 12.0 0.0 0.0 0.0 0.0 76.0 Median 72.0 4.5 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Average 62.5 21.5 2.1 2.0 1.6 1.7 1.1 0.1 0.1 0.2 0.4 2.0 0.0 0.3 0.0 0.0 0.1 0.0 0.0 0.0 0.0 3.9 Respondents 101 74 54 46 34 34 29 7 6 6 6 4 3 2 2 1 1 0 0 0 0 36 with exposure N=102. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: This figure displays the distribution of the actual allocation to the different currencies. It shows the minimum, the second quartile group, the median, the third quartile group, and the maximum for the 102 respondents to the question. The data are divided into four equal-sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The median marks the midpoint of the data and is displayed as the line between the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are smaller than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (lowest point). The horizontal red line refers to the average of each category. Considering only the central banks with exposure, the US dollar and the euro maintain their significant lead in the reserve currency space (Figure 15). More than half of the surveyed central banks had exposure to the British pound, and their average allocation amounted to 3.9 percent. Exposure to CNY accounted for the fourth-highest number of central banks (with an exposure that averages 4.3 percent). A third of respondents also had exposure to the Japanese yen (average share of 5.1 percent) and the Australian dollar (average share of 4.8 percent). Results and Observations  | Central Bank Reserve Management Practices 26 Figure 15. Distribution of allocations to individual currencies for respondents with exposure 100 90 Actual currency allocation (percent) 80 70 60 50 40 30 20 10 0 USD EUR GBP CNY AUD JPY CAD SGD KRW NOK NZD ZAR DKK INR SEK HKD RUB Other Min 1.0 0.1 0.0 0.0 1.0 0.1 0.0 0.0 1.0 1.0 1.0 9.0 1.0 1.0 1.0 1.0 12.0 1.0 Max 100.0 100.0 23.0 20.0 32.0 18.0 13.0 3.0 5.0 6.0 23.0 74.0 1.0 25.0 1.0 1.0 12.0 76.0 Median 73.0 16.0 2.0 2.0 3.0 4.0 3.0 1.0 2.0 2.0 3.0 59.0 1.0 13.0 1.0 1.0 12.0 5.0 Average 63.2 29.6 3.9 4.3 4.8 5.1 3.8 1.2 2.3 2.7 6.0 50.3 1.0 13.0 1.0 1.0 12.0 11.0 Respondents 101 74 54 46 34 34 29 7 6 6 6 4 3 2 2 1 1 36 with exposure N=102. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: This figure displays the distribution of the actual allocation to the different currencies. It shows the minimum, the second quartile group, the median, the third quartile group, and the maximum for the 102 respondents to the question. The data are divided into four equal-sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The median marks the midpoint of the data and is displayed as the line between the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are smaller than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (lowest point). The horizontal red line refers to the average of each category. Analyzing the currency composition of central banks by region, income group, and reserve level, we find that trade and financial flows are important inputs for this allocation decision(Table 2).19 As noted, the average central bank in the survey held 62.5 percent of the portfolio in US dollars and 21.5 percent in euros. However, countries in the Americas and the Caribbean had the greatest average exposure to the US dollar (91 percent). In comparison, countries in Europe had the largest average allocation to the euro (41.6 percent). By income group, the participation of the US dollar was higher in middle- and low-income countries. The share of the euro was greater in high-income countries: this is in keeping with its importance in Ito and McCauley (2019) also show that the currency composition of reserves relates strongly to the 19  co-movement of the domestic currency with key currency and the currency invoicing of trade. Results and Observations  | Central Bank Reserve Management Practices 27 Europe. With regard to reserve levels, the participation of the dollar decreased, and that of the euro increased with increasing size of reserves. These results suggest that countries with smaller reserves and lower-income levels concentrated on the US dollar because of the currency's importance in global trade and financial flows. Table 2. The average allocation to individual currencies by geographic region, country income group, and size Geographic region USD EUR GBP CNY JPY Americas & Caribbean 91.4 2.1 2.1 0.7 0.4 Europe & Central Asia 41.1 41.6 2.4 1.5 3.2 Middle East & Africa 66.8 16.3 1.6 2.7 0.5 South & East Asia and Pacific 68.7 5.1 2.3 3.7 2.4 Country-income group High income 52.7 30.3 3.3 1.2 3.3 Upper middle income 64.4 21.7 1.4 1.6 1.0 Lower middle & low income 72.0 11.4 1.3 3.1 0.5 Size of assets under management (US$) Less than 3 billion 66.3 16.8 1.5 0.9 0.3 3 to 10 billion 64.5 25.0 1.9 1.9 1.1 10 to 50 billion 59.6 24.6 2.5 3.6 3.2 More than 50 billion 58.6 19.9 2.7 1.7 2.7 Foreign exchange regime Not applicable 20.5 50.5 0.0 0.0 0.0 Floating 56.5 25.9 2.8 2.0 3.1 Soft Peg 71.3 13.7 1.2 2.3 0.5 Hard Peg 62.7 32.0 3.8 0.0 0.0 Grand Total 62.5 21.5 2.1 2 1.7 Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Compared to our 2020 survey, the Chinese renminbi became a more critical com- ponent of reserve portfolios while the role of the US dollar declined on average by almost three percent. From 2020 to 2021, the average allocation to the renminbi increased by two percent, while the share of the dollar decreased by slightly more (Figure 16). Other reserve currencies, such as the euro, the British pound, the Australian dollar, and the Japanese yen also decreased as a share of reserve portfo- lios. The interest rate differential between China and other countries may explain Results and Observations  | Central Bank Reserve Management Practices 28 this phenomenon—the Chinese CNY now offers the highest nominal yield among the leading reserve currencies. Countries like Australia and Canada, which were previously attractive for reserve managers, reduced their policy rates to almost zero percent in recent years. At the same time, the increasing importance of CNY may also be a reflection of its inclusion in the SDR basket in 2016. Figure 16. Distribution of changes in major currency allocation (compared to the previous survey)20 80 Change in actual currency allocation (percent) 60 40 20 0 -20 -40 USD EUR GBP CNY JPY AUD CAD Min -46.0 -33.9 -5.6 -5.3 -10.3 -4.7 -2.0 Max 28.0 55.0 4.3 15.0 7.3 2.0 4.0 Median -0.1 -0.1 -0.2 0.2 -0.1 0.0 0.0 Average -1.7 -0.1 -0.8 1.6 -0.8 -0.3 0.5 Number of 72 50 37 27 19 22 16 respondents N=72. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: This figure displays the distribution of the changes in actual allocation to the different currencies compared to our previous survey. It shows the minimum, the second quartile group, the median, the third quartile group, and the maximum for the 72 respondents in common between the two surveys. The data are divided into four equal- sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The median marks the midpoint of the data and is displayed as the line between the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are small- er than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (lowest point). The horizontal red line refers to the average of each category. 20  The currencies displayed in this chart refer to those currencies that at least 50 percent of respondent cen- tral banks identified as eligible for investment. Results and Observations  | Central Bank Reserve Management Practices 29 d. Eligible Asset Classes As in previous RAMP surveys, central banks favored asset classes with high li- quidity and low risk in line with their capital preservation and liquidity objectives (Figure 17). Ninety-five percent of central banks were eligible to invest in bonds; 91 percent in bank deposits; 88 percent in sovereign, supranational, and agency (SSA) securities;21 and 88 percent in money market instruments (see Figure 17). Seven- ty-four percent of the respondents also allow investments in gold and 53 percent in inflation-indexed bonds. Figure 17. Eligible asset classes Government bonds 95% Bank deposits 91% Money market instruments 88% SSA 88% Gold 74% Inflation-indexed bonds 53% Covered bonds 46% IG corporate bonds 41% MBS 37% EM bonds 32% ABS 21% Equity (developed markets) 18% EM equity 8% High-yield corporate bonds 3% Other 21% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of respondents N=113. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: ABS=asset-backed securities, EM=emerging market, IG=investment grade, MBS=mortgage-backed securi- ties, SSA=supranational, sovereign, and agency. Almost three-quarters of central banks were allowed to invest in nontraditional asset classes such as corporate bonds, emerging market bonds, covered bonds, mortgage-backed securities (MBS), and equities. For this survey, we classify as traditional asset classes any that more than 50 percent of the respondent central banks deem to be eligible. As Figure 17 highlights, therefore, bank deposits, gov- ernment bonds, money market instruments, supranational, subsovereign & agen- cies, gold, and inflation-linked bonds belong to the traditional asset classes. At the same time, we categorize the remainder of the asset classes identified in Figure 17 as nontraditional. While these asset classes are typically riskier on a stand-alone basis, they also tend to offer higher expected returns and increase portfolio diver- sification (Hentov et al. 2019). A search for yield amid an environment of low or negative interest rates in developed markets may be driving the interest in riskier asset classes. The preferred nontraditional asset class was covered bonds (46 per- cent of respondents), followed by investment-grade corporate bonds (41 percent), The sovereign securities in the SSA category are typically in foreign currency. When governments issue in 21  their own currency, it is considered a government bond. Results and Observations  | Central Bank Reserve Management Practices 30 mortgage-backed securities (37 percent), emerging market bonds (32 percent), and asset-backed securities (ABS) (21 percent). Finally, equity investing was still rare for central banks: as many as 18 percent of institutions can invest in developed market equities and eight percent in emerging market equities. Compared to our 2020 survey, central banks continued to expand their universe of eligible asset classes to most asset classes within fixed income, reflecting a continued appetite for diversification (Figure 18). Within the traditional asset classes, inflation-linked bonds saw the most significant increase in eligibility—now 61 percent of institutions allow investments in these securities, versus 46 percent in 2020. As regards the nontraditional assets, the eligibility of mortgage-backed securities (MBS) saw the most significant increase (13 percent). More central banks may be interested in MBS to enhance returns, while maintaining high credit quali- ty. Other nontraditional asset classes added to reserve managers' eligible universes are covered bonds (53 percent of central banks in 2021, compared to 40 percent in 2020) and corporate bonds (42 percent of central banks in 2021, compared to 36 percent in 2020). The proportion of central banks that allow equity investments did not change significantly. Figure 18. Changes in asset class eligibility (compared to the previous survey) inflation-indexed bonds 15% MBS 13% Covered bonds 13% Money market instruments 8% IG corporate bonds 6% ABS 5% SSA 5% Bank deposits 5% Gold 2% EM bonds 1% Equity (developed markets) 1% Government bonds 1% 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% Net change in percentage of respondents N=85. Source: Third RAMP survey on the Reserve Management Practices of Central Banks, Second RAMP survey on the Reserve Management Practices of Central Banks. Note: ABS=asset-backed securities, EM=emerging market, IG=investment grade, MBS=mortgage-backed securi- ties, SSA=supranational, sovereign, and agency. Our 2021 survey results highlight that traditional asset classes have the highest allocation in reserve portfolios. The median and average allocation to government bonds stood at 34.1 percent. For the other traditional asset classes, allocations to these asset classes differed significantly across survey respondents because the median and average allocation vary considerably. For example, central banks held an average of 23 percent in bank deposits with a median allocation of 15 percent. Similarly, reserve managers allotted an average of 11.9 percent to SSA securities with a median of nine percent (see Figure 19). Results and Observations  | Central Bank Reserve Management Practices 31 Figure 19. Distribution of allocations to individual asset classes 100 90 Actual asset allocation (percent) 80 70 60 50 40 30 20 10 0 Govern- Bank SSA Money Gold IG Equity EM MBS Covered Inflation- ABS High-yield EM Other ment deposits market corporate (developed bonds bonds indexed corporate equity bonds instruments bonds markets) bonds bonds Min 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Max 87.0 100.0 71.0 100 76.0 50.0 35.0 15.0 22.0 48.0 11.0 14.0 8.0 10.0 77.0 Median 34.0 15.0 9.0 3.5 2.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Average 34.1 23.0 11.9 10.2 8.0 2.4 1.7 1.2 1.0 1.3 0.4 0.3 0.2 0.1 4.2 Respondents 80 72 69 59 56 28 14 21 17 24 15 5 3 3 24 with exposure N=92. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: This figure displays the distribution of the actual allocation to the different asset classes. It shows the min- imum, the second quartile group, the median, the third quartile group, and the maximum for the 92 respondents to the question. The data are divided into four equal-sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The median marks the midpoint of the data and is displayed as the line between the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are smaller than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (lowest point). The horizontal red line refers to the average of each category. Note: ABS=asset-backed securities, EM=emerging market, IG=investment grade, MBS=mortgage-backed securi- ties, SSA=supranational, sovereign, and agency. Despite the increasing endeavor of central banks globally to include nontradi- tional asset classes in reserve portfolios, on average, these asset classes still only make up a small portion of reserve portfolios, with an average allocation of 8.2 percent across all respondent banks. On average, reserve managers held about 80 percent of their reserve in four asset classes: government bonds (34.1 percent), bank deposits (23 percent), supranational (11.9 percent), and money market instru- ments (10.2 percent). Gold is also a reserve asset, with an average allocation of 8 percent. In terms of nontraditional asset classes, investment-grade corporates had the highest average allocation across central banks (2.4 percent), followed by devel- oped market equities (1.7 percent) and emerging market bonds (1.2 percent). Furthermore, the allocations to each nontraditional asset class are small, limiting the potential return enhancement that these asset classes could provide. The av- erage individual share of these types of assets is below 12 percent of total reserves. Results and Observations  | Central Bank Reserve Management Practices 32 Figure 20 shows the reported asset allocations of respondents' portfolios and pres- ents only data for the institutions that indicated exposure to a specific asset class. (It does not, therefore, reflect the impact of central banks who did not report an allocation.) For each financial instrument, the Figure displays the range of institu- tions' reported shares and quartiles, as well as the median and average. In most cases, the median allocation to nontraditional asset classes was below 10 percent. Figure 20 provides the number of central banks that invest in this specific asset class. Breaking down respondent central banks' exposure to individual non- traditional asset classes shows that the highest number of central banks allocated to investment-grade corporates. Thirty percent of respondent central banks had an average exposure of nearly eight percent of total reserves. Twenty-three percent of respondents invested in emerging market bonds with an average allocation of 5 percent. Eighteen percent of respondents had investments in mortgage-backed securities, and 15 percent in developed market equities, with an average allocation of 5.2 and 11.4 percent, respectively. Figure 20. Distribution of the allocation to individual asset classes for respondents with exposure 100 90 Actual asset allocation (percent) 80 70 60 50 40 30 20 10 0 Govern- Bank SSA Money Gold IG Covered EM Inflation- MBS Equity ABS High-yield EM Other ment deposits market corporate bonds bonds indexed (developed corporate equity bonds instruments bonds bonds markets) bonds Min 3.0 1.0 1.0 1.0 0.1 0.5 0.9 0.7 0.0 0.1 1.0 1.0 4.0 0.7 1.0 Max 87.0 100.0 71.0 100.0 76.0 50.0 48.0 15.0 11.0 22.0 35.0 14.0 8.0 10.0 77.0 Median 38.5 22.5 12.0 10.0 6.0 5.5 2.2 3.0 1.0 2.3 9.5 5.8 5.0 2.0 11.5 Average 39.2 29.3 15.9 15.9 13.1 7.9 5.0 5.0 2.7 5.2 11.4 6.0 5.7 4.2 16.1 Respondents 80 72 69 59 56 28 24 21 15 17 14 5 3 3 24 with exposure N=92. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: This figure displays the distribution of the actual allocation to the different asset classes. It shows the min- imum, the second quartile group, the median, the third quartile group, and the maximum for the 92 respondents to the question. The data are divided into four equal-sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The median marks the midpoint of the data and is displayed as the line between the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are smaller than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (lowest point). The horizontal red line refers to the average of each category. Note: ABS=asset-backed securities, EM=emerging market, IG=investment grade, MBS=mortgage-backed securi- ties, SSA=supranational, sovereign, and agency. Results and Observations  | Central Bank Reserve Management Practices 33 The analysis of asset allocations of reserve portfolios by reserve level, region, and income group reveals several notable patterns (Table 3). As discussed above, cen- tral banks held an average allocation of 34.1 percent in government bonds. They also had an average allocation of 23 percent in bank deposits, twelve percent in sovereign, supranational, and agency (SSA) bonds, and ten percent in money mar- ket instruments. These average allocations notably change if we compare reserve portfolios across regions. Central banks in Europe and Central Asia exhibited the largest allocation to nontraditional assets. At the same time, South and East Asia and Pacific institutions invested the most on average in traditional asset classes. Respondents in the Americas and the Caribbean had relatively larger allocations to government bonds but smaller allocations to other asset classes. In Europe and Central Asia, the average participation of government bonds, SSA, and gold was relatively large. Nevertheless, exposure to bank deposits and money market prod- ucts was more muted, probably due to the negative rates in Europe. In the Middle East and Africa, central banks had the smallest average allocation to government bonds and the largest average allocation to bank deposits while maintaining a con- siderable exposure to SSA. Finally, countries in the Pacific and South and East Asia had relatively large average allocations to all traditional asset classes, especially government bonds. By income level, central banks of high-income countries invested the most in non- traditional asset classes, with an average allocation of 13.8 percent. Lower-middle- and low-income countries, in turn, present the lowest allocation to nontraditional asset classes, with a 3.6 percent allocation on average, while investing the most in traditional assets. In terms of the specific asset classes, high-income central banks had reserve port- folios concentrated in government bonds. By contrast, lower-middle- and low-in- come countries had the largest allocation in bank deposits. Comparing the composition of reserve portfolios across the absolute size of re- serves, we find that central banks that manage less than $3 billion had a rather large allocation to bank deposits, probably indicating a less sophisticated invest- ment framework. At the same time, institutions with more than $50 billion of re- serves had a higher share of their reserves invested in government bonds and only limited exposure to deposits, which may be the result of the difficulties that tend to arise when investing large portfolios in less liquid instruments. Finally, in terms of the foreign exchange regime of the respondents, we observe that countries with a floating regime invested the most in nontraditional asset classes, with an allocation of 10.8 percent on average (compared to only eight per- cent for the sample as a whole). Results and Observations  | Central Bank Reserve Management Practices 34 Table 3. Average allocation to individual asset classes by geographic region, country income group, and reserve size22 MONEY TRADITIONAL GOVERNMENT BANK NONTRADITIONAL SSA MARKET GOLD ASSET BONDS DEPOSITS ASSET CLASSES Geographic region INSTRUMENTS CLASSES Americas & Caribbean 43.2 19.1 12.8 8.7 3.2 88.2 7.8 Europe & Central Asia 37.1 12.0 12.7 7.5 14.4 83.9 10.6 Middle East & Africa 26.1 37.3 9.4 11.4 5.1 89.6 6.4 South & East Asia and Pacific 31.5 21.8 15.5 19.4 4.3 92.9 6.2 Country income group High income 41.7 8.9 11.9 9.4 10.0 82.5 13.8 Upper middle income 32.0 21.0 17.4 11.8 8.4 91.2 6.5 Lower middle & low income 27.7 38.8 8.2 10.0 5.6 90.5 3.6 Size of assets under management (US$) Less than 3 billion 25.1 34.0 11.8 12.2 4.3 87.5 8.7 3 to 10 billion 31.7 27.8 12.6 7.5 4.7 84.4 9.4 10 to 50 billion 35.8 20.3 11.5 13.9 11.4 93.3 5.2 More than 50 45.4 7.4 11.6 7.9 12.9 86.2 9.0 billion Foreign exchange regime Not applicable 8.0 92.0 0.0 0.0 0.0 100.0 0.0 Floating 43.9 10.0 11.8 9.1 8.8 84.4 10.8 Soft Peg 22.5 36.5 12.8 12.1 7.3 91.2 6.0 Hard Peg 42.1 18.3 9.1 7.9 8.1 85.5 5.2 Grand Total 34.1 23 11.9 10.2 8 87.6 8.2 Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: SSA=supranational, sovereign, and agency. Comparing this year's survey results with those of 2020, we note that institutions slightly shifted the makeup of their traditional assets. On average, reserve manag- ers reduced the allocation to bank deposits and money market instruments. At the same time, central banks also altered the composition of their nontraditional assets as investment-grade corporate bonds saw an average one percent decline in alloca- tion. The average share of bank deposits and money market products decreased by 0.4 and 2.5 percent, respectively. In parallel, the average allocation to government bonds grew slightly (Figure 21) by an average of 0.3 percent. This rebalancing may The classification of traditional and nontraditional asset classes excludes the asset type “other”. 22  Results and Observations  | Central Bank Reserve Management Practices 35 be a reaction to the challenge of investing in short-term products in the current yield environment and the need to improve expected returns. Finally, another notable change compared to the 2020 survey is the average increase of allocation to gold, though this may simply reflect a 20 percent increase in the price of bullion from the end of 2018 through to the end of 2020. Figure 21. Changes in asset allocation (compared to the previous survey) 80 70 Change in actual asset allocation (percent) 60 50 40 30 20 10 0 -10 -20 -30 -40 -50 -60 -70 -80 Govern- SSA Bank Money Gold Covered EM IG MBS Inflation- Equity ABS EM High-yield Other ment deposits market bonds bonds corporate indexed (developed equity corporate bonds instru- bonds bonds markets) bonds ments Min -58.2 -69.6 -51.0 -71.0 -15.2 -9.0 -11.0 -21.0 -7.1 -6.1 -6.3 -5.9 -0.4 0.0 -18.0 Max 50.1 53.8 69.7 68.0 16.3 6.0 15.0 10.8 10.0 2.2 8.4 6.5 10.0 5.0 62.0 Median 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Average 0.3 1.0 -0.4 -2.5 1.3 0.0 0.4 -0.9 0.2 -0.1 0.2 0.1 0.1 0.1 2.6 Number of 61 58 55 53 36 19 18 18 15 14 7 5 3 2 30 respondents N=65. Source: Third RAMP survey on the Reserve Management Practices of Central Banks, Second RAMP survey on the Reserve Management Practices of Central Banks. Note: This figure displays the distribution of the changes in actual allocation to the different asset classes com- pared to our previous survey. It shows the minimum, the second quartile group, the median, the third quartile group, and the maximum for the 65 respondents in common between the two surveys. The data are divided into four equal-sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The median marks the midpoint of the data and is displayed as the line between the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are smaller than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (low- est point). The horizontal red line refers to the average of each category. Note: ABS=asset-backed securities, EM=emerging market, IG=investment grade, MBS=mortgage-backed securi- ties, SSA=supranational, sovereign, and agency. Results and Observations  | Central Bank Reserve Management Practices 36 This survey again confirmed the results of our previous two surveys on central bank risk-taking and reserve adequacy. We cannot find a clear relationship between the level of risk central banks take in their reserve management operations as proxied by their allocation to nontraditional assets with their import coverage as a proxy for a central bank's reserve adequacy (Figure 22). Figure 22. Import coverage and allocation to nontraditional asset classes 15 Import Coverage (months) 10 5 0 0 5 10 15 20 25 30 35 40 45 50 55 60 Allocation to nontraditional asset classes (percent) N=48. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Another notable observation is that central banks rarely include gold in their SAA optimization framework. The third RAMP survey attempted to gain further in- sights on central bank gold holdings. Therefore, the survey included a question on how central banks treat gold in their strategic asset allocation exercise. We found that 76 percent of central banks had gold in their foreign reserves. At the same time, only 20 percent of those institutions include gold in their SAA optimization framework, suggesting that the allocation to gold in central bank reserve portfolios is rarely decided on the basis of risk and return considerations or correlations with other asset classes. In several countries, historical precedents or the requirement to buy gold from local producers may determine the level of gold holdings, explain- ing why this asset is not included in the optimization framework. 3.3 PORTFOLIO MANAGEMENT a. Active management An active investment approach is one of the investment styles that central banks use to manage reserve portfolios. In the absence of a benchmark, portfolio man- agers often buy securities and hold them until they mature. For institutions with Results and Observations  | Central Bank Reserve Management Practices 37 a consistent SAA framework, the choice is typically between passive and active management. Passively managed portfolios attempt to replicate the composition of their benchmark. On the other hand, actively managed portfolios deviate from the benchmark in pursuit of a higher return. An active investment strategy will always be more resource-intensive than a passive strategy because it requires investment management skills in both the portfolio and the risk management teams. However, central banks can deploy various strategies for active management. An arguably less intensive, and more risk-averse approach is an enhanced indexing strategy that permits only limited deviations from the benchmark. The survey found that most respondents (almost 80 percent) managed their port- folios actively. For institutions that deploy a tranching framework, the investment style deployed by institutions varies by tranche (Figure 23). As the investment horizon increased, respondent central banks appeared to have a higher tolerance for active risk in their portfolios. We find that 56 percent of respondents use either active management or enhanced indexing for their liquidity portfolio, and this per- centage increases slightly to 60 percent for the investment tranche. Figure 23. Portfolio management style by tranche Total portfolio Liquidity tranche Investment tranche Total portfolio (untranched) 50% 40% 37% 37% Percentage of respondents 33% 31% 31% 30% 28% 28% 25% 23% 22% 22% 20% 20% 19% 18% 13% 12% 10% 0% Active Enha- Passive Buy Active Enha- Passive Buy Active Enha- Passive Buy Active Enha- Passive Buy nced and nced and nced and nced and Indexing Hold Indexing Hold Indexing Hold Indexing Hold N=116. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Most central banks use an active management style, but many have a separate risk metric to limit active risk-taking. This survey inquired about the maximum Results and Observations  | Central Bank Reserve Management Practices 38 tracking error or active risk central banks deploy for active management at the subportfolio level.23 Tracking error limits are a good indicator of the extent of active risk-taking. Only 27 percent of institutions with active risk use tracking error limits to manage their active risk-taking. Of these institutions, over half of the respon- dents had tracking error limits below 100 basis points (Figure 24), whereas 41 per- cent utilized limits of 100 basis points or more. Figure 24. Maximum tracking error for actively managed portfolios or portfolios that apply enhanced indexing 21% 100 basis points and above 41% 51-99 basis points 17% 31-50 basis points 10-30 basis points 21% N=24. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Reserve managers can take active risk by deviating from their benchmark cur- rency composition. Our 2021 survey results show that reserve managers did not take much active currency risk (Figure 25). Generally, most central banks tended to keep the currency composition of their portfolios close to their benchmarks—the median difference between the portfolio and the benchmark for all currencies is zero percent, with a trivial dispersion for most participants. Nonetheless, two cur- rency allocations stood out. Central banks allocated an average 0.6 percent higher share of their reserves to the US dollar than their strategic asset allocation. The op- posite was true for the euro—the central banks had on average a 0.6 percent lower allocation to that currency than their strategic asset allocation. Tracking error is the standard deviation of expected returns relative to the benchmark. It is one of the most 23  common metrics used to assess the level of active risk in a portfolio. Results and Observations  | Central Bank Reserve Management Practices 39 Figure 25. Currency deviations from the benchmark 40 35 30 Di erence in actual and strategic currency allocation (percent) 25 20 15 10 5 0 -5 -10 -15 -20 -25 USD EUR GBP JPY AUD CNY CAD Min -24.0 -13.0 -15.0 -7.0 -13.0 -12.0 -10.0 Max 39.0 11.0 23.0 9.0 2.0 13.0 2.0 Median 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Average 0.6 -0.6 -0.3 0.0 -0.4 -0.1 -0.4 Number of 44 38 25 21 17 17 13 respondents N=78. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: This figure displays the distribution of the difference in actual and strategic currency allocation. It shows the minimum, the second quartile group, the median, the third quartile group, and the maximum for 78 respondents. The data are divided into four equal-sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The median marks the midpoint of the data and is displayed as the line be- tween the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are smaller than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (lowest point). The horizontal red line refers to the average of each category. Regarding active deviations from the strategic asset allocation, our latest survey indicates that reserve managers took significant positions away from their strate- gic asset allocation weights. In our 2021 survey, reserve managers revealed consid- erable overweight in government bonds and underweight in SSA securities (Figure 26). On average, reserve portfolios had an allocation to government bonds that was 15.2 percent above their benchmarks. The median overallocation to government bonds for all respondents amounted to seven percent, indicating that many central banks hold these overweight positions. In contrast, on average, central banks were underweight in SSA securities by six percent, which may be driven by a heightened focus on liquidity and the expectation of spread widening. Results and Observations  | Central Bank Reserve Management Practices 40 Figure 26. Asset class deviations from the benchmark 100 80 Di erence in actual and strategic currency allocation (percent) 60 40 20 0 -20 -40 -60 Govern- SSA Bank Money Covered Gold IG MBS EM Inflation- ABS Equity High- EM Other ment deposits market bonds corporate bonds indexed (developed yield equity bonds instru- bonds bonds markets) corporate ments bonds Min -51.0 -71.0 -51.0 -15.0 -8.0 -35.0 -31.0 -14.0 -3.0 -1.0 -14.0 -2.0 -2.0 0.0 -60.0 Max 100.0 85.0 75.0 57.0 11.0 9.0 12.0 30.0 16.0 4.0 1.0 1.0 0.0 0.0 4.0 Median 7.0 -3.0 -1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Average 15.2 -5.8 -1.0 -0.9 0.0 -1.0 -1.0 0.0 0.1 0.1 -0.3 0.0 0.0 0.0 -1.9 Number of 50 43 39 34 14 13 12 9 8 7 4 2 1 0.0 10 respondents N=90. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: This figure displays the distribution of the difference between actual and strategic asset allocation for the different asset classes. It shows the minimum, the second quartile group, the median, the third quartile group, and the maximum for 90 respondents. The data are divided into four equal-sized quartiles made up of 25 percent of the data. The lines dividing the groups (2nd and 3rd) are the quartiles. The median marks the midpoint of the data and is displayed as the line between the 2nd and the 3rd quartile groups. Half of the data are either greater than or equal to this value, while the remaining 50 percent are smaller than this value. The upper and bottom whiskers represent the maximum (highest point) and minimum (lowest point). The horizontal red line refers to the average of each category. Despite the widespread use of active investment management styles and strat- egies, the practice of active risk budgeting is still limited at central banks. Risk budgeting compares active strategies using a similar metric (tracking error) to find an efficient active risk allocation. Best practice in investment management holds that reserve managers should build a robust risk budgeting framework to deploy active risk efficiently. Only 27 percent of organizations that took active risk system- atically allocated tracking error across active strategies or asset classes. Using an active risk budgeting methodology becomes particularly important when institu- tions use active investment management strategies and deviate significantly from their strategic currency and asset composition. Several factors may explain the limited adoption of active risk budgeting strategies by central banks, not least the likelihood that institutions may have insufficient resources and skillsets to design a proper risk budgeting allocation framework (see Section 3.4). Results and Observations  | Central Bank Reserve Management Practices 41 b. Derivatives, Investment Instruments, and Investment Strategies As institutional investors increase risk and diversification, the use of derivatives, investment instruments, and strategies tends to rise. Derivatives allow investors to hedge risks, facilitate the implementation of new asset classes, or even increase expected returns by giving more flexibility to portfolio managers. These properties are particularly beneficial for more complex portfolios. Additionally, investment instruments like Exchange Traded Funds (ETF) simplify the access to nontradi- tional asset classes at a low cost. Finally, investment strategies such as repurchase agreements (repos) and securities lending allow institutions to generate additional returns. Our survey found that central banks deployed a limited number of derivatives, investment instruments, and investment strategies (Figure 27). More than 50 per- cent of reserve managers used foreign exchange forwards, swaps, and interest rate futures. As for investment strategies, more than half of institutions have authorized repurchase agreements and securities lending. Other central banks may not have adopted these alternative instruments and strategies because they may not be permitted to use them. Even when the use of these instruments is not debarred, in- stitutions may lack the complex legal documentation entailed. Meanwhile, institu- tions may have only a weak framework for collateral management—or none at all— or may in any case be wary about taking on the counterparty credit risk embedded in these transactions. Figure 27. Use of derivatives, investment instruments, and investment strategies 100% 3% 7% 6% 9% 10% 11% 9% 15% 16% 7% 90% 7% 9% 8% 9% 25% 11% Percentage of respondents 80% 10% 15% 15% 54% 53% 70% 58% 59% 58% 17% 60% 50% 77% 80% 76% 72% 73% 69% 8% 40% 63% 13% 64% 9% 67% 9% 13% 46% 30% 23% 26% 20% 24% 28% 22% 10% 9% 9% 11% 10% 10% 9% 10% 10% 12% 10% 12% 5% 7% 7% 6% 0% Asset swaps Credit default swaps Currency futures Currency options ETFs FX options FX swaps Gold options Gold swaps Interest rate futures Interest rate options Interest rate swaps Repos Securities lending FX forwards Yes Under consideration No, not authorized No, appropriate legal agreements are not in place (such as ISDAs) N=112. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Results and Observations  | Central Bank Reserve Management Practices 42 The use of most derivatives increased from 2018 to 2021. Foreign exchange and interest rate swaps are the derivatives that have attracted the most attention—an additional 16 and 10 percent of central banks have included these instruments respectively in their investment guidelines in the past three years (Figure 28). Also, more central banks started using currency options (seven percent increase), in- terest rate options (six percent increase), and asset swaps (six percent increase). However, central banks do not necessarily trade derivatives frequently or in con- siderable amounts even if they are eligible to do so. Some of the obstacles to trad- ing these instruments more actively may include the lack of legal documentation, limited dealer capacity, and accounting complexities. Figure 28. Net change between surveys in the percentage of central banks that use each derivative FX swap 16% Interest Rate swaps 10% Currency options 7% Interest rate options 6% Asset swaps 6% Currency futures 4% Gold swaps 3% FX forwards 3% Credit default swaps 3% Interest rate futures 1% Gold options 0% FX options 0% 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% Net change in percentage of respondents N=70. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. c. External Management Although central banks continue to hire external managers, these tend, as previ- ously, to manage only a small portion of reserve assets. We observe that 75 percent of respondents have at least one external manager (Figure 29). At the same time, there is significant dispersion in the share of the overall reserve portfolio delegat- ed to third parties. Thirty-seven percent of central banks with external managers allocate less than five percent of reserve assets to external management programs, Results and Observations  | Central Bank Reserve Management Practices 43 while only 14 percent of respondents delegate more than 30 percent.24 The latter group of central banks had an average of 51 percent of their reserves managed externally. From 2018 to 2021, the number of central banks that used one or more external managers rose from 69 to 72 percent of all respondents. Figure 29. Share of institutions’ reserve assets externally managed as of December 31, 2020 More than 30% 0% - Do not use 14% external managers 25% < S ≤ 30% 26% 6% 20% < S ≤ 25% 3% 9% 10% < S ≤ 15% 7% 27% 10% < S ≤ 15% 9% 0% < S ≤ 5% 5% < S ≤ 10% N=115. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Knowledge transfer and return enhancement are the main motivations when hir- ing external managers. More than 70 percent of respondents cite these factors as highly relevant to their decision to hire external managers. A significant proportion of institutions also cite internal constraints (45 percent) and increased access to investment strategies (41 percent) (Figure 30). Most central banks with external asset management programs work with fewer than five financial institu- 24  tions (77 percent) (Figure 29). However, there are significant differences in the average number of external managers by income level, region, and absolute size of reserve. Central banks in high-income and up- per-middle-income countries are more likely to have six or more external managers (27 percent of central banks in each of those income groups). By region, central banks in the Pacific and South and East Asia are more likely to have six or more external managers (30 percent of institutions in those regions). Also, respondents with higher reserve levels are more likely to have many more managers: 53 percent of central banks with more than $50 billion in reserves have six or more managers, compared to only five percent for those with less than $3 billion. Results and Observations  | Central Bank Reserve Management Practices 44 Figure 30. Relevance of considerations when opting to hire external managers to manage part of reserve assets (percentage of respondents with external managers) Highly relevant Somewhat relevant Not relevant Capacity building by knowledge transfer 75% 15% 9% Enhancement of 68% 20% 8% return/performance Internal constraints (e.g., 45% 35% 13% management skills) Increased access to resource-in 41% 39% 9% tensive investment strategies Benchmark for 18% 39% 31% internal sta Environment, social, and corporate 11% 28% 48% governance (ESG) considerations Other 4% 1% 24% 0% 50% 100% 0% 50% 100% 0% 50% 100% Percentage of respondents N=104. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. A significant number of central banks use external managers to get exposure to nontraditional asset classes. Most institutions allow external managers to invest in traditional asset classes such as government and SSA bonds. However, more than a third of respondents with external managers allow investment-grade corporate bonds, mortgage-backed securities, and covered bonds. A lower proportion of central banks uses external managers to invest in emerging market bonds, as- set-backed securities, and equity (Figure 31). Results and Observations  | Central Bank Reserve Management Practices 45 Figure 31. Asset classes used by external managers Government bonds 85% Supranational, sub-sovereign, 79% agency (SSA) Money market instruments 59% Bank deposits 56% Inflation-indexed bonds 48% Covered bonds 42% Investment-grade 40% corporate bonds Mortgage-backed securities 36% Emerging market bonds 28% Asset-backed securities 19% Equity 18% Emerging market equity 11% Other 8% Gold 7% High-yield corporate bonds 7% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% Net change in percentage of respondents N=85. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. 3.4 RISK MANAGEMENT Managing risks is an essential component of the reserve management process. A robust risk management framework identifies and quantifies the most relevant risks to help institutions maintain them within predefined levels (International Monetary Fund 2014). Credit and market risk management are critical in these ex- traordinary times, and the survey had questions on these areas. Additionally, man- aging active risk arising from active investment strategies is a crucial component of a central bank's risk management framework. Finally, reporting relevant risk from reserve management operations is also a part of decision-making for central bank reserve managers. Industry practice suggests that reserve managers' risk management has to evolve as they invest in a broader set of asset classes, move down the credit spectrum, or take on significant active risk. Table 4 describes the risk management tools available to appropriately manage risk and exposure to different asset classes on a stand-alone basis. In other words, it assesses whether a portfolio that only has that asset class should use a specific risk management methodology. In our analysis of central banks' risk management operations, we use this framework to assess risk management in central banks. Results and Observations  | Central Bank Reserve Management Practices 46 Table 4. Recommended (minimum) risk management tools by asset class CREDIT RISK MARKET RISK Market indica- Fundamental Quantitative mod- Internal model Maximum Credit ratings tors (e.g. bond credit analysis els provided by Expected that combines currency devi- Tracking error ASSET CLASS from rating spreads, CDS some of the in- (e.g., debt to third parties (e.g., Duration limit VaR limit shortfall ations from the (TE) limit Sector limits agencies spreads, equity equity, interest Moody’s EDF, (CVaR) limit dicators above benchmark prices) coverage) Bloomberg SRSK) Yes (different Yes for com- Yes, if currency Yes for commercial from maturity Yes for com- Bank deposits Yes bank deposits mercial bank limit to manage deviations are mercial banks deposits allowed liquidity risk) Yes for commercial Yes for commer- Yes, if currency Yes for com- Money market Yes bank and corporate cial bank and Yes deviations are mercial banks instruments issuers corporate issuers allowed and corporates Yes, if investing Yes, if investing Yes, if currency Yes, if active Yes for issuers Government bonds Yes for issuers with Yes for issuers Desirable for issuers in securities with in securities with Yes Yes deviations are management is with lower (developed markets) lower ratings with lower ratings with lower ratings maturities above maturities above allowed allowed ratings one year one year Yes, if investing Yes, if investing Yes, if currency Yes, if active Yes for issuers Yes for issuers with Yes for issuers Desirable for issuers in securities with in securities with Inflation-linked bonds Yes lower ratings with lower ratings with lower ratings Yes maturities above maturities above deviations are management is with lower allowed allowed ratings one year one year Yes, if investing Yes, if investing Supranational, Yes, if currency Yes, if active Yes for issuers Yes for issuers with Yes for issuers Desirable for issuers in securities with in securities with sovereign, and Yes lower ratings with lower ratings with lower ratings Yes maturities above maturities above deviations are management is with lower agency (SSA) allowed allowed ratings one year one year Yes, if investing Yes, if investing Yes, if currency Yes, if active Yes for issuers Yes for issuers with Yes for issuers in securities with in securities with Covered bonds Yes lower ratings with lower ratings Yes maturities above maturities above deviations are management is with lower allowed allowed ratings one year one year Agency mort- Yes, if currency Yes, if active gage-backed Yes Yes Yes deviations are management is Yes securities allowed allowed Results and Observations  | Central Bank Reserve Management Practices 47 CREDIT RISK MARKET RISK Market indica- Fundamental Quantitative mod- Internal model Maximum Credit ratings tors (e.g. bond credit analysis els provided by Expected that combines currency devi- Tracking error ASSET CLASS from rating spreads, CDS some of the in- (e.g., debt to third parties (e.g., Duration limit VaR limit shortfall ations from the (TE) limit Sector limits agencies spreads, equity equity, interest Moody’s EDF, (CVaR) limit dicators above benchmark prices) coverage) Bloomberg SRSK) Yes for instru- Yes, if currency Yes, if active Asset-backed Yes for instruments Yes ments with lower Yes Yes Yes deviations are management is Yes securities with lower ratings ratings allowed allowed Yes, if currency Yes, if active Investment-grade Yes Yes Desirable Yes Desirable Yes Yes Yes deviations are management is Yes corporate bonds allowed allowed Yes, if currency Yes, if active High-yield corporate Yes Yes Desirable Yes Desirable Yes Yes Yes deviations are management is Yes bonds allowed allowed Yes, if currency Yes, if active Equity (developed Yes Yes deviations are management is Yes markets) allowed allowed Yes, if currency Yes, if active Emerging market Yes Yes Desirable Yes Desirable Yes Yes Yes deviations are management is Yes bonds allowed allowed Yes, if currency Yes, if active Emerging market Yes Yes deviations are management is Yes equity allowed allowed Yes, if active Yes for gold Gold deposits Yes Yes management is Yes allowed Source: Authors’ assessments. Note: Depending on how these limits are established, some of them could be nonbinding (e.g., an institution only has a CVaR limit but also monitors duration and VaR). Results and Observations  | Central Bank Reserve Management Practices 48 a. Credit Risk Management Central banks managed credit risk by mainly investing in debt securities with high credit ratings (Figure 32). Credit risk is especially concerning for central banks because defaults may impact their reputation. As seen in the previous sur- veys, most survey participants had minimum ratings in the investment-grade cat- egory (BBB+/BBB/BBB- or above), mainly in line with their investment objective of safety and liquidity. For all asset classes, the most frequent minimum credit rating was A+/A/A-, confirming the continued conservative stance of central banks. Figure 32. Minimum credit rating for eligible issuers in the asset classes listed Government bonds Supranational/agency bonds A+/A/A- 42% A+/A/A- 44% AA+/AA/AA- 16% AA+/AA/AA- 17% AAA 3% AAA 8% BB+ and below 1% BB+ and below 0% BBB+/BBB/BBB- 34% BBB+/BBB/BBB- 26% Not applicable 4% Not applicable 5% 0% 10% 20% 30% 40% 50% 0% 10% 20% 30% 40% 50% % of respondents % of respondents Corporate bonds Commercial banks (for deposits) A+/A/A- 28% A+/A/A- 48% AA+/AA/AA- 7% AA+/AA/AA- 8% AAA 1% AAA 0% BB+ and below 1% BB+ and below 0% BBB+/BBB/BBB- 18% BBB+/BBB/BBB- 19% Not applicable 45% Not applicable 24% 0% 10% 20% 30% 40% 50% 0% 10% 20% 30% 40% 50% % of respondents % of respondents N=114, N=111, N=106, and N=108, respectively. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Compared to our 2020 survey, most central banks kept their minimum credit ratings unchanged for most asset classes, although some loosened their ratings for government bonds, SSA bonds, and commercial bank deposits (Figure 33). Notably, 17 percent of central banks reduced their minimum ratings for govern- ment bonds and 11 percent for SSA bonds, typically by one notch. At the same time, 10 percent of the institutions tightened their minimum rating requirements for government bonds. There was a marginal increase in minimum credit ratings of commercial bank deposits—11 percent of the respondents tightened them, and only six percent loosened them. Results and Observations  | Central Bank Reserve Management Practices 49 Figure 33. Number of central banks with changes in credit ratings from the 2020 survey 47% Corporates 6% 3% 59% Commercial banks 6% 11% 74% SSA 11% 11% 67% Government bonds 17% 10% 0% 10% 20% 30% 40% 50% 60% 70% 80% Percentage of respondents Tightened Loosened No Change N=72, N=80, N=84, N=79, respectively. Source: Third RAMP survey on the Reserve Management Practices of Central Banks, Second RAMP survey on the Reserve Management Practices of Central Banks. Rating agencies remain the primary source of information on credit risk (Figure 34). Since the previous survey, the use of methodologies for credit risk analysis had not changed significantly. Almost all central banks that responded to our survey used credit ratings in their risk management process. Nevertheless, although all respondents that invested in corporate bonds use credit ratings, they do not rely exclusively on these ratings, opting instead to complement them with other meth- odologies (Figure 35); 72 percent use market indicators, such as bond spreads, CDS spreads, or equity prices, while internal credit risk models to assess credit risk are less common (24 percent). For central banks with minimum credit ratings of BBB-/ BBB/BBB+, internal credit risk models play a more critical role. Forty-four percent of institutions in that group use them (see Figure 36), the same percentage as the complete sample. Figure 34. Credit risk assessment methodologies used for issuers and counterparties Credit ratings from rating agencies 93% Market indicators (e.g., bond spreads, credit default swaps (CDS), equity prices) 56% Internal model that combines 44% some of the indicators above Fundamental credit analysis (e.g., debt or equity, interest coverage) 35% Quantitative models provided by third parties (e.g., Moody's 15% expected default frequency (EDF)) Other 9% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of respondents N=117. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Results and Observations  | Central Bank Reserve Management Practices 50 Figure 35. Credit assessment methodologies for respondents with corporate credit exposure Credit ratings from rating agencies 100% Market indicators (e.g., bond spreads, credit default swaps (CDS), equity prices) 72% Fundamental credit analysis (e.g., debt or equity, interest coverage) 38% Quantitative models provided by third parties (e.g., Moody's 28% expected default frequency (EDF)) Internal model that combines 24% some of the indicators above Other 7% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of respondents N=29. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Figure 36. Credit assessment methodologies for respondents with minimum credit rating of BBB-/BBB/BBB+ Credit ratings from rating agencies 96% Market indicators (e.g., bond spreads, credit default swaps (CDS), equity prices) 64% Internal model that combines 44% some of the indicators above Fundamental credit analysis (e.g., debt or equity, interest coverage) 40% Quantitative models provided by third parties (e.g., Moody's 18% expected default frequency (EDF)) Other 4% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of respondents N=45. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. A significant number of central banks that take credit risk in various portfolios do not produce aggregate credit risk measures. Only 47 percent of respondents use a portfolio credit risk model to calculate numbers such as expected and unexpect- ed loss, credit value at risk (VaR), and expected credit shortfall (conditional credit value at risk). Most institutions that produce these metrics (72 percent) use an in- house or custom credit risk model. Basel II- or III-type models or those provided by vendors are not frequently used, with 19 percent of central banks using Basel III- Results and Observations  | Central Bank Reserve Management Practices 51 type models, and 21 percent using vendor models. However, compared to our 2020 survey, institutions have made some progress in using in-house risk models. The share of central banks deploying that methodology increased from 49 to 60 percent of total respondents from 2020 to 2021 (Figure 37). Figure 37. Credit risk model used In-house or custom credit risk model 72% Vendor credit risk models (e.g., CreditMetrics, Moody's KMV, CreditRisk+) 21% Basel II/III model 19% Other 4% 0% 10% 20% 30% 40% 50% 60% 70% 80% Percentage of respondents N=53. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. b. Market Risk Management Central banks utilized multiple metrics to manage market risk in their reserve management operations. Market risk refers to the risk of losses arising from move- ments in market prices. As most reserve portfolios are concentrated in high-grade fixed-income instruments, duration is the most essential market risk measure. Eighty-three percent of institutions have limits on portfolio duration to manage duration risk (Figure 38).25 Forty-three percent of central banks with multiple currency portfolios deployed limits on the maximum currency deviation from the benchmark to manage their tactical currency risk. Nevertheless, 57 percent of central banks did not. While they could choose to manage the currency risk through other quantitative risk measures such as VaR limit, only 12 percent of central banks with multiple curren- cy exposures do so. However, most respondents do not deviate from the currency composition of the benchmark. For central banks with a diversified asset composition, capturing the risk of these portfolios beyond simple duration limits is highly desirable. Probabilistic risk limits are critical tools for central banks to consider because of the various market risks embedded in their reserve portfolios. For example, instruments with embed- ded options such as MBS, ABS, and callable bonds have exposure to factors that cannot be captured by employing simple duration limits (such as volatility, spreads changes, or inflation). Best practice suggests that VaR and CVaR limits are appro- priate tools to manage the risks embedded in these asset classes. Although these limits are often included in the strategic asset allocation process, it is remarkable that only a third of respondent central banks with diversified asset class portfolios Duration risk refers to the sensitivity of fixed income instruments prices to changes in yields. 25  Results and Observations  | Central Bank Reserve Management Practices 52 have adopted these types of probabilistic risk tools to measure and limit the overall market risk at the individual and total reserve portfolios. Comparing our 2021 survey results with those of 2020, we do not find any signifi- cant changes despite central banks’ efforts to build more diversified asset portfoli- os, suggesting that there continues to be room for improvement. Figure 38. Limits used to manage market risk 100% 90% 83% 80% Percentage of respondents 70% 60% 50% 48% 40% 38% 34% 31% 30% 21% 20% 10% 10% 4% 0% Duration Maximum Tracking Sector VaR (CVaR) Other None limit currency error (TE) limits limit limit deviations limit from the benchmark N=118. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. c. Active risk Only a small portion of central banks deployed limits on tracking error to manage the active risk of their portfolios. As mentioned above, 80 percent of central banks employ an active style when managing their portfolio. According to industry best practice, one important tool to manage the active risk is to use tracking error lim- its. But only 27 percent of respondents that adopt an active management style put in place such limits to control the active risk of their portfolio; this reveals scope for the enhancement of practices. d. Reporting and Transparency Reserve managers reported on a variety of risk measures to boards and invest- ment committees at regular intervals. Boards and investment committees typically approve changes to the reserve management investment policies and guidelines (Anasashvili et al. 2020). The board and the investment committee regularly receive reports on exposures and risk metrics at the individual and reserve portfolio levels to oversee reserve management activities. Several survey questions asked Results and Observations  | Central Bank Reserve Management Practices 53 respondents to specify what type of risk measures they report to the board and investment committee and how frequently. The investment committee usually received more information on risk and perfor- mance than the governor or the board, but reports often lacked some key metrics (Table 5). The investment committee usually has more oversight over the day- to-day management of the portfolios, and it is reasonable that they receive more comprehensive reports. The investment committee reports of more than half of central banks do include the most relevant risk and performance metrics. Howev- er, most central banks do not report the tracking error, the credit VaR, or the rating breakdown to the investment committee, the governor, or the board. International experience suggests that comprehensive risk and performance reports—albeit at reasonable intervals and uncluttered with excessive detail—are critical for boards and investment committees to properly manage and monitor reserve management activities. An empirical study that looks at governance structures and reporting also suggests that central banks where the head of risk management directly re- ports to the board hold more diversified currency and asset portfolios, controlling for reserve adequacy and the overall institutional environment. The anchoring of risk at the board level may further allow reserve managers to run riskier reserve portfolios controlling for reserve adequacy and the macroenvironment as well as for the broader institutional environment (Klingebiel et al. 2021). Regarding the frequency of such reporting, 45 percent of respondents reported to their boards quarterly. In comparison, reserve managers issued reports to the gov- ernor and the investment committee monthly (41 and 49 percent respectively). Table 5. Risk and performance indicators reported to the governor, board, or investment committee (percentage of respondents) Absolute Excess Tracking Duration Market Credit Rating Other return return error VaR/CVaR VaR/CVaR breakdown Governor 76% 67% 36% 71% 50% 18% 46% 7% Board 79% 66% 35% 67% 48% 20% 47% 8% Investment 86% 76% 48% 86% 62% 24% 58% 6% committee N=117. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. ENVIRONMENTAL, SOCIAL, AND 3.5.  GOVERNANCE (ESG) Including Environmental, Social, and Governance (ESG) factors in the investment analysis is becoming more common. Climate change and the social challenges Results and Observations  | Central Bank Reserve Management Practices 54 arising from the COVID-19 pandemic have raised the institutional investor’s aware- ness of their impact on the environment and society. As a result, global sustainable investment grew 15 percent between 2018 and 2020, reaching $35 trillion (Global Sustainable Investment Alliance 2020). Reserve managers are also participating in this trend, although the adoption of ESG investing is still low. Twenty-six percent of respondents included ESG objec- tives in their investment policy or investment framework (Figure 39).26 Most insti- tutions with such a policy or framework applied these strategies to their internally managed portfolios (94 percent), while 52 percent also used them in their external- ly managed portfolios. Moreover, almost a third of central banks are considering ESG, signaling that the adoption of this investment style may grow in the future. However, nearly half of respondents were neither implementing nor considering the adoption of ESG in their investment operations. The low adoption rate of ESG in reserve portfolios is attributable to the fact that central banks concentrate on high-quality fixed-income assets, where the availability of ESG instruments and strategies is limited (see Bouyé, Klingebiel, and Ruiz 2021). Figure 39. Inclusion of environmental, social, and governance (ESG) factors in the investment policy 17% Yes 45% 9% No, but it is included in the investment framework No, but it is is under discussion No 28% N=117. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Having a positive impact and maintaining their reputation are the main drivers of ESG implementation in central banks. Almost three-quarters of central banks im- plementing ESG allude to impact and reputation as motivating factors (Figure 40). Compared with the 2019 survey, the number of institutions that have included ESG in the reserve portfolio 26  has not changed. The 2019 survey had one question on ESG adoption. In contrast, the 2021 survey changed the wording of that question, including references to the investment policy and framework, and included new questions to understand the motivations, the approaches, and the allocations to those strategies. Some of the questions included specific definitions to facilitate interpretations (e.g., negative screening or green bonds). In view of those changes, we do not attempt to compare the two surveys when direct comparison is no longer possible. Results and Observations  | Central Bank Reserve Management Practices 55 Central banks are crucial entities in their countries, and, leading by example, they may promote ESG adoption in both the private and the public sectors. Interestingly, improving the risk-return profile of the portfolios or responding to public or peer pressure is not essential for central banks, despite typically being cited as highly relevant factors for other institutional investors. Figure 40. Motivating factors to incorporate ESG (institutions with ESG in their investment policy or framework) 100% 90% 78% 80% 73% Percentage of respondents 70% 60% 50% 40% 36% 30% 20% 13% 13% 10% 7% 0% Positive Reputation/ Risk/return Peer Public Other environmental brand enhance- pressure pressure or social ment impact N=45. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. ESG implementation varied by region, income level, and level of reserves. The breakdown by region shows that South and East Asia and the Pacific are the re- gions that led on ESG implementation (45 percent of institutions in those regions), followed by Europe and Central Asia (40 percent).27 Another interesting observation is that ESG adoption increases with income level because 42 percent of central banks in high-income countries have implemented ESG, compared with 28 percent in upper-middle-income and 12 percent in lower-middle and low-income coun- tries. Finally, central banks with more significant reserve levels were more likely to execute this strategy—58 percent of institutions with more than $50 billion in reserves do so, significantly above those with $10–50 billion (36 percent), $3–10 billion (3 percent), or less than $3 billion (12 percent). Reserve portfolios that include equities were also more likely to have ESG in their investment process. Fifty-five percent of central banks invested in equities 27  In the Americas and the Caribbean 14 percent of central banks have implemented ESG, compared to seven percent in the Middle East and Africa. Results and Observations  | Central Bank Reserve Management Practices 56 also adopted ESG. However, only 22 percent of institutions without equities did so, again highlighting that the availability of investment strategies in high-grade fixed income strategies is a constraining factor (Bouyé, Klingebiel, and Ruiz 2021). Impact investing through green, social, and sustainability bonds was the pre- ferred ESG strategy in central banks (Table 6).28 Two-thirds of central banks that had implemented ESG, either in the investment policy or investment framework, bought thematic bonds issued by SSA. Approximately a third of these institutions also bought labeled corporate bonds. The second most prevalent approach was negative screening, applied mainly to corporate bonds (45 percent of central banks with ESG), government bonds (32 percent), and equity (32 percent). A third of the respondents in this group also claimed to use ESG integration to invest in corporate bonds. Other ESG approaches were not adopted at a significant frequency. Table 6. ESG strategies used (percentage of institutions with ESG) Government/ Corporate High-yield Equity SSA bonds bonds bonds Negative/exclusionary 32% 45% 10% 32% screening Positive/best-in-class 16% 16% 0% 6% screening ESG integration 23% 35% 10% 16% Impact investing (i.e., green bonds, social Not 68% 35% 3% bonds, sustainability applicable bonds) Impact investing differ- ent from green, social, 6% 3% 0% 3% or sustainability bonds Active ownership 3% 10% 3% 19% and engagement N=31. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Nevertheless, the participation of green, social, and sustainability bonds in port- folios is minimal. Eighty-two percent of central banks with green bonds issued by governments and SSA have one percent or less of their portfolios invested in these instruments. Only two central banks have an investment above 10 percent of the portfolio. Fewer respondents buy social or sustainability bonds; even if they do, exposures are much smaller. As highlighted in Bouyé, Klingebiel, and Ruiz (2021), In this document, green, social, and sustainability bonds are also referred to as thematic or labeled bonds. 28  Results and Observations  | Central Bank Reserve Management Practices 57 although the market capitalization of these types of bonds is growing significantly, the overall market size for these instruments is still negligible. This presents a sig- nificant obstacle for any central bank seeking to increase its allocation here.  ANDEMIC BUSINESS CONTINUITY 3.6. P MANAGEMENT The COVID-19 pandemic placed an enormous operational burden on most insti- tutions. To conduct their unprecedented monetary policy and business operations seamlessly during the pandemic, many central banks implemented a set of busi- ness continuity or crisis management measures. This year's survey included several questions on central banks' business continuity practices. From an operational perspective, around half of central banks had not prepared for a pandemic before 2020. COVID-19 was an unprecedented event insofar as ep- idemics in recent decades had neither spread globally nor resulted in generalized lockdowns. Consequently, 48 percent of respondents did not have business conti- nuity protocols or arrangements in place for a pandemic scenario in 2019 (Figure 41). Thirty-one percent of respondents had protocols or arrangements in place for a pandemic but had to change them in 2020 as COVID-19 cases grew worldwide. Only 20 percent of central banks had a pandemic business continuity framework in place that did not need any adjustment. Figure 41. Did your institution have business continuity protocols and arrangements in place for a pandemic scenario as of December 31, 2019? 1% Yes 19% 20% Yes, but we adjusted the protocol in 2020 No, but a general business continuity or crisis management framework was in place No, but the general business continuity 29% 31% or crisis management framework was adjusted in 2020 Other N=117. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Although most institutions changed their work arrangements in response to the pandemic, fewer than half of the central banks moved to a full working-from- home mode. As COVID-19 emerged in various countries, most central banks im- Results and Observations  | Central Bank Reserve Management Practices 58 plemented a rotational schedule, where staff would work from the office on certain days or weeks and work from home at other times (Figure 42). Nearly 40 percent of central banks conducted full remote operations (working from home). Very few central banks continued to operate from their offices exclusively. Figure 42. Work arrangements adopted in response to the pandemic Rotational schedule (i.e., sta working from the o ce on select 77% days or weeks on a rotational basis) Dispersed sta across several sites to reduce occupancy rate at the o ce 35% and ensure social distancing Full remote operation (i.e., all 35% sta working from home) Continued to operate from the o ce (i.e., all sta 6% working from the o ce) Other 9% 0% 10% 20% 30% 40% 50% 60% 70% 80% Percentage of respondents N=116. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: Respondents could choose more than one answer. Therefore, results do not add up to 100 percent. For many central banks, working remotely proved difficult as staff could not per- form all of the critical functions outside the office, mainly back-office operations. Approximately half of the central banks could only perform settlement and pay- ment operations and custody management from the office (Figure 43). At the same time, two-thirds of respondents could fully execute trades remotely, and more than 80 percent of respondents could perform critical functions, such as reporting or compliance from home. Only seven percent of respondents indicated that they could execute any of these functions exclusively in the office. Figure 43. Reserve management activities that can be performed remotely Internal reporting on performance and risk 86% Monitoring risk metrics 83% Monitoring compliance with investment guidelines 82% Trading 70% Custodian management 52% Settlement and payment (i.e., SWIFT operations) 47% None 7% Other 2% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% Percentage of respondents N=112. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Results and Observations  | Central Bank Reserve Management Practices 59 Central banks cited cybersecurity and a lack of appropriate hardware and soft- ware as the main constraints to perform activities remotely (Figure 44). Most respondents indicated that they have access to secure networks, reliable internet connections at home, and access to vendor services. Interestingly, senior man- agement concerns about remote work or the lack of human resource policies for flexible work arrangements did not seem to be a significant constraint. Figure 44. Main constraints to remote working General cybersecurity concerns or 42% cybersecurity policy restrictions Lack of appropriate hardware and software (e.g., No access to the portfolio management/straight-through 20% processing (STP) systems from home) No access to vendor services (e.g., Bloomberg, risk management solutions) 16% No access to a secure 15% network (e.g., via VPN) Senior management concerns with respect to sta working remotely 12% Unreliable internet connection at home 11% Lack of flexible work arrangements per human resources (HR) policies 4% Other 2% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% Percentage of respondents N=105. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. 3.7. AUDIT Auditing reserve management activities is critical for central banks. An indepen- dent internal audit should regularly evaluate procedures and controls to monitor compliance with the institution's policies and articulate proposals to the board or senior management to improve the governance and organizational structure (International Monetary Fund 2014). For most central banks, reserve management plays a critical role in the institution's financial situation, if not the most important. As a result, auditors review the processes deployed in reserve management, usu- ally during the annual audit of the financial statements. Auditors are the so-called "third line of defense" in reserve management, after those that execute the oper- ations (front office) and those that monitor risk and compliance (middle office). Auditors assess and assure that the governance framework is effective and policy and process are in place. Results and Observations  | Central Bank Reserve Management Practices 60 The survey found that auditors play three essential roles. The first and most important is to perform an independent assessment of internal controls, risk management, and governance (93 percent of respondents) (Figure 45). The second role is to assure the effectiveness of reserve management systems and processes (69 percent). Finally, auditors monitor the risk of critical operational areas such as trading, settlements, and accounting (65 percent). Therefore, an effective auditing function is essential for a well-performing reserve management process. Figure 45. Internal audit of reserve management operations: roles To assess internal controls, risk management, and governance 93% independently and objectively To assure the e ectiveness of reserve management systems and processes 69% To monitor the risk of key operational areas such as trading,settlements, and 65% accounting as a third line of defense To o er advisory services to reserve management 12% Other 3% 0% 20% 40% 60% 80% 100% Percentage of respondents N=118. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Ensuring compliance with the regulatory framework and policies and reviewing trade processes and controls are the main activities of internal audit. More than three-quarters of internal audit teams at respondent central banks executed these activities (Figure 46). For a considerable number of respondents (47 percent), audi- tors also performed sample trade controls. However, the analysis by income level reveals notable differences. For high-income countries, the essential activities of their internal audit function were sample trade control (76 percent of institutions in this group), followed by acting as a trusted advisor (64 percent). For middle- and low-income countries, internal auditors were focused on ensuring compliance with policies and the regulatory framework (89 percent for upper-middle-income and 78 for lower-middle and low-income countries) and the review of trade processes and controls (76 percent for upper-middle-income and 68 percent for lower-middle and low-income countries). Results and Observations  | Central Bank Reserve Management Practices 61 Figure 46. Internal audit of reserve management operations: activities Ensures compliance with regulatory framework and policy 78% Trade process review and controls 72% Sample trade control 47% Trade controls 32% Portfolio-level controls 29% Trusted advisor 9% Other 5% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% Percentage of respondents N=116. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Recognized external auditing firms are frequently in charge of auditing financial statements. More than 80 percent of surveyed central banks had this arrangement. Internal audit also performed this function in approximately half of the central banks (Figure 47), but more frequently in upper-middle-income countries (71 per- cent) and the Americas and the Caribbean (68 percent). However, given the unique role of central banks, for 68 percent of respondents, both internal and external auditors reviewed the financial statements. Figure 47. Responsibility for auditing financial statements Recognized external auditing firms 88% Internal audit 48% Local bank regulator or government supervision entities 14% Other 8% 0% 20% 40% 60% 80% 100% Percentage of respondents N=118. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Results and Observations  | Central Bank Reserve Management Practices 62 3.8. THE NUMERAIRE AND PUBLIC DISCLO- SURE POLICIES OF CENTRAL BANKS The US dollar remains the most frequently used numeraire for reporting purpos- es, followed by the domestic currency (Figure 48). More than 60 percent of central banks reported performance in US dollars and 35 percent in local currency. Using the SDR or another currency basket was unusual. A breakdown by region shows that countries in the Americas, the Caribbean, the Middle East, and Africa de- ployed the US dollar as numeraire more regularly (90 percent of respondents in the Americas and the Caribbean and 79 percent of those in the Middle East and Africa). European and Central Asian central banks utilized either the domestic currency (43 percent) or the euro (43 percent).29 Institutions in South and East Asia and the Pacific preferred the US dollar (68 percent) and frequently used the domestic cur- rency (50 percent). Figure 48. Currencies used for performance reporting as a numeraire USD 62% Domestic currency 35% EUR 20% Currency basket 7% SDR 4% Other 4% 0% 10% 20% 30% 40% 50% 60% 70% Percentage of respondents N=118. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Although most institutions produce internal reports, disclosing extensive in- formation on reserve management activities to the public was unusual (Figure 49). The only data that most central banks (60 percent) revealed was the currency composition of reserves. A significant proportion also published information on eligible asset classes (50 percent), performance (44 percent), and asset allocations (41 percent). However, central banks did not generally share with the public the main guiding principles of the investment policy or granular information on re- serve management practices. Fewer than a quarter of central banks divulged their investment policies, guidelines, risk metrics, or the characteristics of external man- agement programs. Results and Observations  | Central Bank Reserve Management Practices 63 Figure 49. Information disclosed to the public, either on a mandatory or voluntary basis (percentage of respondents) 100% 41% 41% 41% 41% 41% 41% 41% 41% 10% 25% 25% 24% 21% 80% Percentage of respondents 44% 41% 60% 50% 60% 52% 78% 67% 67% 40% 64% 47% 51% 59% 59% 44% 59% 59% 59% 59% 59% 59% 59% 20% 33% 23% 7% 6% 9% 8% 10% 9% 11% 11% 0% Currency Eligible Performance Asset External Risk metrics Investment Investment Portfolio composition asset (e.g., absolute alloca- managers (e.g., (e.g., policies horizon bench- classes and relative tions asset classes duration, and marks returns over an and proportion tracking guidelines investment of assets exter- error, and horizon) nally managed) VaR) YES NO NOT APPLICABLE N=116. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. 3.9. ACCOUNTING PRACTICES FOR GOLD Accounting practices for gold vary considerably. As explained in section 3.2, this survey included questions on gold. The most common approach to report returns on gold was to account for unrealized gains and losses as reserves, either in equity or a non-equity account (43 percent). For 36 percent of institutions, the mark-to- market valuation of gold affected the profit and loss account. Notably, few institu- tions (18 percent) include unrealized gains and losses on gold in calculating distri- butions to their governments (Figure 50). Figure 50. Accounting practices for unrealized gains and losses on gold Not reported Other Reserves in equity account 7% 15% 25% Reserves in non-equity account 18% 18% Profit and loss (P&L) account impacting distributions to 18% the government Profit and loss (P&L) account without impacting distributions to the government N=85. Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Results and Observations  | Central Bank Reserve Management Practices 64  oncluding 4. C Observations and Policy Implications The COVID-19 pandemic upended reserve management operations in central banks. These institutions had to respond simultaneously to three significant shocks. Some countries liquidated part of their reserve holdings to cope with in- creasing liquidity needs in foreign currency. Additionally, lockdowns forced institu- tions to redesign operations as their staff had to work from home. Finally, the pol- icy response in developed markets caused a significant decrease in interest rates, improving the return on reserves in 2020 but worsening it for 2021 and beyond. We conducted the third RAMP survey during the COVID-19 pandemic. We found several striking changes between the 2020 and the 2021 surveys as institutions re- sponded to this challenging environment. Central banks put more weight on safety and liquidity than on return to respond to the potentially higher reserve need. Furthermore, they adjusted their operations and were able to execute a significant number of functions remotely. Finally, reserve managers decreased portfolio du- ration and expanded eligibility and exposure to more currencies and asset classes, going beyond the traditional choices. Central banks confronted the crisis, and managed through it: in so doing, they innovated in certain areas. Investment operations added derivatives and instru- ments. Furthermore, more central banks engaged with external managers. They also explored how to implement ESG in their portfolios. Nonetheless, as identified in the two previous RAMP surveys, risk management and transparency have not evolved on a par with asset allocation and portfolio manage- ment. The survey results highlight that reserve managers do not measure and man- age credit and market risk with state-of-the-art tools. Enhancing risk management practices is particularly important for institutions with exposure to nontraditional assets. Most institutions also remain reluctant to improve the transparency of their reserve management practices. Improving practices in these areas will be vital if the art and science of reserve management are to evolve further. Concluding Observations and Policy Implications  | Central Bank Reserve Management Practices 65 References Aizenman, J. A. 2007. "International Reserves: Precautionary versus Mercantilist Views, Theory and Evidence." Open Economies Review 18 (2): 191–214. Alekasir, K., N. Anasashvili, M. Antonio, P. Dongsoo Hong, D. Klingebiel, G. Pereira Alves. 2019. Inaugural RAMP Survey on the Reserve Management Practices of Central Banks: Results and Observations. RAMP (Reserve Advisory and Management Partnership). Washington, DC, USA: The World Bank Group. Anasashvili, N., P. Dongsoo Hong, D. Klingebiel, C. Herrero Montes, and M. Ruiz Gil. 2020. Central Bank Reserve Management Practices: Insights into Public Asset Management from the Second RAMP Survey. RAMP (Reserve Advisory and Management Partnership). Washington, DC, USA: The World Bank Group. Bouyé, E., D. Klingebiel, and M. Ruiz Gil. 2021. Environmental, Social, and Governance Invest- ing: A Primer for Central Banks' Reserve Managers. Washington, DC, USA: The World Bank Group. Brinson, G. P., L. R. Hood, and G. L. Beebower. 1986. "Determinants of Portfolio Perfor- mance." Financial Analysts Journal 42 (4): 39–44. DOI: 10.2469/faj.v42.n4.39. Emmer, S., M. Kratz, and D. Tasche. 2015. "What Is the Best Risk Measure in Practice? A Comparison of Standard Measures." Journal of Risk 18 (2): 31–60. European Central Bank, International Relations Committee Task Force. 2006. "The Accumu- lation of Foreign Reserves." Occasional Paper Series No. 43. European Central Bank, Frankfurt, Germany. Grinold, R., and R. Kahn. 1999. Active Portfolio Management: A Quantitative Approach for Producing Superior Returns and Controlling Risk. McGraw-Hill Education. New York. Habib, M.M., L. Stracca, and F. Venditti. 2020. "The Fundamentals of Safe Assets". Working Paper Series, European Central Bank (ECB). No. 2355. ECB, Frankfurt, Germany Heller, H. R. 1966. "Optimal International Reserves." The Economic Journal 76 (302): 296–311. Hentov E., A. Petrov, D. Kyriakopoulou, and P. Ortlieb. 2019. "How Do Central Banks Invest? Embracing Risk in Official Reserves." State Street Global Advisors and OMFIF, Bos- ton, MA. Ibbotson, R. 2010. "The Importance of Asset Allocation." Financial Analysts Journal 66 (2). DOI: 10.2307/27809175. International Monetary Fund. 2014. Revised Guidelines for Foreign Exchange Reserve Manage- ment. Washington, DC: International Monetary Fund. Ito, H., and R. McCauley. 2019. "The Currency Composition of Foreign Exchange Reserves." BIS Working Papers 828. Basel, Switzerland: Bank for International Settlements. Jeanne, O., and R. Rancière. 2011. "The Optimal Level of International Reserves for Emerg- References  | Central Bank Reserve Management Practices 66 ing Market Countries: A New Formula and Some Applications." The Economic Jour- nal 121 (155): 905–30. Lütkebohmert, E. 2009. Concentration Risk in Credit Portfolios. Berlin: Springer. Klingebiel, D., C. Herrero Montes, M. Ruiz Gil, and J. Seward. 2021. Central Bank Governance and Reserve Portfolios Investment Policies: An Empirical Analysis. Forthcoming World Bank Policy Research Working Papers Series. Washington, DC, USA: The World Bank Group. Siegel, M. 2008. Stocks for the Long Run: The Definitive Guide to Financial Market Returns and Long-Term Investment Strategies. New York: McGraw Hill. Wang, L. A. 2019. "Determinants of Currency Composition of Reserves: A Portfolio Theory Approach with an Application to RMB." International Monetary Fund (IMF) Work- ing Paper 19 (52). IMF, Washington, DC. World Bank Group. 2019. World Bank Data. Retrieved from new country classifications by income level: 2017–2018: https://datahelpdesk.worldbank.org/knowledgebase/arti- cles/906519-world-bank-country-and-lending-groups. References  | Central Bank Reserve Management Practices 67 Appendix Table A.1. Use of Tranching Geographic region TRANCHING NO TRANCHING Americas & Caribbean 82.6% 17.4% Europe & Central Asia 51.2% 48.8% Middle East & Africa 90.6% 9.4% South & East Asia and Pacific 77.3% 22.7% Country-income group High income 54.6% 45.5% Upper middle income 79.4% 20.6% Lower middle & low income 87.5% 12.5% Size of assets under management (US$) Less than 3 billion 72.4% 27.6% 3 to 10 billion 90.3% 9.7% 10 to 50 billion 72.0% 28.0% More than 50 billion 57.6% 42.4% Foreign exchange regime Not applicable 100.0% Floating 60.0% 40.0% Soft Peg 83.7% 16.3% Hard Peg 100.0% Grand Total 72.9% 27.1% Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Appendix  | Central Bank Reserve Management Practices 68 Table A.2. Average Investment Horizon of the Tranches (months) TOTAL LIQUIDITY INVESTMENT TOTAL Geographic region PORTFOLIO TRANCHE TRANCHE PORTFOLIO (UNTRANCHED) Americas & Caribbean 9.6 49.1 21.5 31.7 Europe & Central Asia 13.6 31.3 12.4 43.1 Middle East & Africa 10.7 43.2 26.2 15.3 South & East Asia and Pacific 18.8 39.0 66.0 22.0 Country-income group High income 20.6 45.5 30.3 46.2 Upper middle income 9.5 38.9 20.5 18.7 Lower middle & low income 10.2 42.0 20.8 27.4 Size of assets under management (US$) Less than 3 billion 10.2 49.0 16.3 35.0 3 to 10 billion 11.4 41.4 23.6 9.7 10 to 50 billion 7.5 31.8 12.0 19.7 More than 50 billion 23.7 41.4 37.6 56.0 Foreign exchange regime Not applicable 7.0 60.0 Floating 12.0 32.5 17.9 40.9 Soft Peg 9.8 43.1 23.6 25.1 Hard Peg 33.6 73.2 66.0 Grand Total 15.6 52.2 35.8 33.0 Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Appendix  | Central Bank Reserve Management Practices 69 Table A.3. Average Duration of the Tranches (months) TOTAL LIQUIDITY INVESTMENT TOTAL Geographic region PORTFOLIO TRANCHE TRANCHE PORTFOLIO (UNTRANCHED) Americas & Caribbean 7.1 23.8 19.6 39.5 Europe & Central Asia 12.5 25.3 14.2 21.3 Middle East & Africa 7.2 25.7 19.9 11.7 South & East Asia and Pacific 6.9 29.6 6.0 21.0 Country-income group High income 14.1 29.0 19.2 25.8 Upper middle income 7.4 28.5 17.9 12.5 Lower middle & low income 5.4 22.1 11.5 22.4 Size of assets under management (US$) Less than 3 billion 8.3 26.7 16.0 20.6 3 to 10 billion 7.2 22.6 20.2 5.7 10 to 50 billion 5.4 28.0 15.6 15.9 More than 50 billion 14.8 29.3 14.2 31.5 Foreign exchange regime Not applicable 2.0 19.5 4.0 Floating 10.0 22.8 15.5 25.0 Soft Peg 6.3 28.5 18.8 15.9 Hard Peg 13.0 28.8 24.0 Grand Total 7.8 24.9 15.6 20.4 Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Appendix  | Central Bank Reserve Management Practices 70 Table A.4. Eligibility by Currency (percentage of respondents within group) Geographic region USD EUR GBP JPY CNY AUD CAD CHF NOK SEK Other NZD DKK SGD KRW HKD ZAR MXN INR TRY BRL RUB Americas & Caribbean 100% 83% 74% 52% 43% 48% 57% 43% 39% 39% 26% 39% 30% 30% 17% 13% Europe & Central Asia 97% 82% 74% 67% 59% 62% 56% 38% 28% 21% 36% 18% 23% 15% 13% 5% 3% 5% 3% 5% 3% 5% Middle East & Africa 100% 94% 84% 63% 75% 56% 59% 56% 38% 44% 31% 28% 31% 16% 25% 13% 28% 16% 9% 9% 9% 3% South & East Asia and 100% 100% 95% 84% 84% 79% 58% 42% 32% 32% 32% 47% 32% 42% 37% 37% 11% 21% 21% 5% 5% 11% Pacific Country-income group High income 100% 81% 74% 67% 60% 62% 60% 36% 40% 36% 17% 36% 36% 19% 21% 14% 10% 12% 10% 10% 7% 5% Upper middle income 97% 97% 94% 73% 67% 76% 67% 58% 39% 42% 42% 42% 33% 36% 27% 24% 3% 9% 3% 3% Lower middle & 100% 89% 76% 58% 68% 45% 47% 45% 21% 21% 39% 13% 16% 16% 16% 5% 18% 8% 8% 5% 5% 5% low income Size of assets under management (US$) Less than 3 billion 100% 93% 69% 38% 41% 34% 34% 28% 10% 7% 31% 14% 3% 14% 3% 14% 3% 3 to 10 billion 97% 90% 83% 67% 63% 60% 60% 53% 40% 43% 33% 37% 37% 27% 27% 13% 17% 20% 10% 13% 10% 7% 10 to 50 billion 100% 88% 88% 76% 72% 64% 60% 44% 28% 32% 28% 24% 24% 12% 16% 4% 4% More than 50 billion 100% 83% 83% 83% 83% 83% 76% 55% 55% 48% 34% 45% 48% 38% 38% 38% 10% 17% 14% 7% 7% 7% Foreign exchange regime Not applicable 100% 100% 50% 50% Floating 100% 88% 82% 79% 79% 72% 68% 51% 44% 40% 28% 39% 39% 32% 32% 21% 9% 14% 7% 7% 5% 2% Soft Peg 100% 91% 83% 57% 57% 53% 51% 40% 23% 26% 36% 21% 17% 15% 11% 6% 13% 6% 6% 4% 4% 6% Hard Peg 86% 71% 57% 29% 14% 29% 29% 43% 29% 29% 29% 29% 29% 14% 14% 14% 14% 14% 14% Grand Total 99% 88% 81% 65% 65% 60% 58% 45% 34% 33% 32% 30% 28% 23% 21% 14% 11% 10% 7% 5% 4% 4% Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Appendix  | Central Bank Reserve Management Practices 71 Table A.5. Eligibility by Asset Class (percentage of respondents within group) MM INFLATION- IG COR- HY COR- GOV. BANK NON- COVERED EM DM EM INSTRU- SSA GOLD INDEXED PORATE MBS ABS OTHER PORATE BONDS DEPOSITS TRADITIONAL BONDS BONDS EQUITY EQUITY Geographic region MENTS BONDS BONDS BONDS Americas & Caribbean 100% 96% 100% 100% 74% 83% 74% 35% 35% 65% 26% 30% 17% 9% Europe & Central Asia 95% 87% 85% 87% 82% 77% 46% 56% 44% 18% 26% 10% 28% 28% 5% 3% Middle East & Africa 88% 97% 84% 81% 66% 69% 50% 41% 34% 41% 34% 28% 22% 9% 9% 6% South & East Asia and 100% 84% 89% 84% 74% 68% 47% 47% 53% 37% 47% 21% 11% 21% 21% Pacific Country-income group High income 100% 78% 93% 93% 68% 85% 59% 51% 54% 41% 37% 22% 20% 37% 15% 5% Upper middle income 94% 100% 91% 85% 70% 76% 61% 52% 33% 42% 33% 24% 24% 9% 6% Lower middle & 90% 97% 82% 85% 85% 62% 41% 36% 33% 28% 26% 18% 21% 5% 3% 3% low income Size of assets under management (US$) Less than 3 billion 90% 97% 83% 86% 62% 69% 45% 48% 41% 31% 17% 24% 28% 14% 7% 3% 3 to 10 billion 97% 97% 90% 87% 87% 73% 57% 47% 43% 37% 27% 30% 20% 17% 3% 3% 10 to 50 billion 92% 84% 92% 88% 84% 84% 48% 48% 32% 32% 36% 4% 12% 4% More than 50 billion 100% 86% 90% 90% 66% 72% 62% 41% 45% 48% 48% 24% 24% 34% 21% 3% Foreign exchange regime Not applicable 100% 100% 100% 100% 50% 100% 50% 50% 50% 50% Floating 100% 84% 95% 93% 69% 81% 62% 52% 52% 43% 41% 22% 22% 29% 12% 3% Soft Peg 87% 98% 81% 81% 81% 64% 45% 36% 30% 28% 23% 19% 19% 6% 4% 2% Hard Peg 100% 100% 83% 83% 83% 83% 33% 67% 33% 50% 17% 17% 33% Grand Total 95% 91% 88% 88% 74% 74% 53% 46% 41% 37% 32% 21% 21% 18% 8% 3% Source: Third RAMP survey on the Reserve Management Practices of Central Banks. Note: ABS=asset-backed securities, DM=developed market, EM=emerging market, Gov.=Government, HY=High-yield, IG=investment grade, MBS=mortgage-backed securities, MM=money market, SSA=supranational, sovereign, and agency. Appendix  | Central Bank Reserve Management Practices 72