## _wp10150

## Source details

**Canonical URL:** [_wp10150](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10150.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10150.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10150.pdf.json)

---

### I. Introduction
- Central banks provide funding to the banking sector mainly through two channels: (1) regular monetary refinancing operations (domestic liquidity) and (2) investing official reserves in short-term bank deposits and securities of foreign banks (cross-border funding).
- Paper objective:
  - Estimate amounts withdrawn by reserve managers from the private sector.
  - Compare withdrawals with central bank support measures.
  - Investigate explanations and implications of procyclical reserve management.
  - Discuss implications for strategic asset allocation.

### II. Reserve management objectives and governance
- Reserves are held primarily for “self-insurance” and to enhance credibility of monetary policy; typically invested in liquid, short-duration government bills and bonds in reserve currencies.
- IMF’s Guidelines for Foreign Exchange Reserve Management (2001) state reserve management should ensure:
  - (1) adequate foreign exchange reserves for a defined range of objectives;
  - (2) liquidity, market, and credit risks are controlled prudently;
  - (3) subject to liquidity and other risk constraints, reasonable earnings generated over the medium to long term.
- “Reserve management function” covers the front and middle office responsibilities inside the central bank; ownership structures (e.g., ministry of finance) can complicate linkages.
- Governance issues noted:
  - Frameworks often impose upper bounds but typically no lower bounds.
  - Decision incentives can favor avoiding losses and reputational risk, producing procyclical behavior.
  - Remedies suggested: realign incentives, introduce passive reserve management for certain asset classes, higher-level decision processes, and international coordination when cross-border externalities exist.

### III. Pre-crisis trends in reserve management
- Major trend: increased focus on return and shift out of government bonds into agency securities, deposits, and sometimes equities starting around the turn of the century.
- Key statistics:
  - Global reserves (excluding gold) rose from US$1.8 trillion in 1998 (year-end) to US$8.5 trillion 11 years later.
  - Share of U.S. dollar in global reserves remained around 60 percent over 2003–09; euro around 30 percent (COFER sample ~60 percent of global reserves).
- Drivers of diversification:
  - Rapid reserve accumulation and treatment of surplus reserves as national wealth.
  - Low interest rates and global search for yield.
  - Perception of limited supply of short-term government debt in reserve currencies.
  - Professionalization of reserve management and benchmarking/peer pressure.
- Investment beliefs and risks:
  - Assumption that investment grade credit spreads adequately compensated risk and ability to hold to maturity.
  - Mean-variance approaches suggested negative marginal risk contribution of investment grade credit, encouraging credit exposure.

### IV. Reserve management during the crisis — behavior and scale
A. Survey and high-level findings
- RMT 2009 survey:
  - 76 percent of responding central banks reduced credit limits in response to the crisis.
  - 85 percent reduced maximum maturity of deposits.
  - Many reduced securities lending.
  - RMT summary: “Central banks joined the flight to quality, cutting deposit holdings, tightening credit limits for assets, reducing exposure to counterparties and emphasizing traditional assets, notably top-rated government bonds and gold.”
- Motivations: primarily fear of losses rather than immediate need to use reserves; many central banks favored holding to maturity.

B. Magnitude and timing of withdrawals (empirical evidence)
- Data sources: IMF SDDS reserves template (63 countries plus ECB/Eurosystem reporting at end-2009), TIC, BIS International Banking Statistics, central bank websites.
- Bank deposit dynamics (SDDS subscribers):
  - In the two months following Lehman Brothers’ collapse an amount equivalent to almost US$170 billion was pulled out of deposits.
  - Between July 2007 and March 2009 the share of commercial bank deposits in reserves of SDDS subscribers dropped 12 percentage points, corresponding to an absolute amount of over US$350 billion.
  - Distribution across central banks:
    - Only 15 percent of central banks that held deposits did not reduce allocation meaningfully.
    - 43 percent reduced share of bank deposits by more than 20 percentage points from peak to trough between July 2007 and March 2009.
    - Over two-thirds saw declines of more than 10 percentage points.
    - Nearly 80 percent of the sample reduced their deposits by more than half (measured by min/max approach with peak preceding trough).
- Change in deposits vs. change in reserves:
  - Several central banks reduced deposits far more than their change in overall reserves; some saw reserves increase while pulling out of deposits.
- Repurchase agreements (reverse repos):
  - From December 2007 some deposit decline was offset by increased net repos until Lehman, when net repos fell sharply (US$124 billion from end-September to end-December 2008).
  - Over July 2007 – March 2009 net repos declined by over US$40 billion in the dataset.
- Aggregate measures and conservative estimates:
  - BIS International Banking Statistics: cross-border liabilities of reporting banks to “official monetary authorities” fell by over US$420 billion from June 2007 to March 2009 (peaked US$1,442 billion in Dec 2007, fell to US$935 billion in Mar 2009).
  - SDDS shows slight increase (US$68 billion between July 2007 and March 2009) in deposits with other central banks and BIS; BIS decreased exposure to banking sector by SDR 92 billion (~over US$140 billion) between March 2008 and March 2009.
  - Conservative aggregate estimate: pullback by central banks and BIS of more than the equivalent of US$500 billion during the crisis, majority after Lehman Brothers’ collapse; most redeployed into government bills and bonds, smaller share deposited with reserve-issuing central banks.
- Agencies and official sector holdings:
  - TIC data: official holdings of U.S. agency debt increased more than tenfold from March 2000 to June 2008; peaked at US$966 billion in June 2008.
  - Between July 2008 and March 2009 official sector reduced holdings of U.S. agency securities by over US$93 billion.

C. Procyclicality of credit-limit frameworks
- Many frameworks rely on external ratings; limits inversely related to creditworthiness, producing inherent procyclicality.
- Back-of-the-envelope estimate: between peak end-2007 and end-2008, changes in aggregate credit limits may explain a reduction of bank exposure of around 10 percent (based on ECB framework applied conservatively to a sample of ten largest banks).
- Given actual pullback around 50 percent, much of the reduction driven by active and semi-active risk reduction rather than automatic limit triggers.

### V. Liquidity support provided by central banks
- Central banks expanded measures: larger volumes, longer maturities, foreign currency liquidity, relaxed collateral, and asset purchases (e.g., Fed purchases of agency debt, Eurosystem purchases of covered bonds).
- Empirical patterns:
  - Monthly changes in central bank claims on domestic banks (IFS database, 176 central banks): modest measures until late 2007; spikes after Bear Stearns near-collapse; largest surge in the two months following Lehman Brothers’ collapse.
  - Concentration: about 93 percent of first month’s increase in claims after Lehman was supplied by Eurosystem central banks and the Fed.
- Currency and scale:
  - Net dollar amount outstanding through dollar swaps peaked at nearly US$600 billion at end-2008.
  - Estimated at least US$1.8 trillion of additional liquidity injected into the banking sector by central banks (mainly Fed and Eurosystem) over the four months following Lehman Brothers’ collapse—excluding later direct purchases of private and government securities.
- Interaction with reserve withdrawals:
  - Some injections were needed to offset funds withdrawn by other central banks (at least US$500 billion).
  - Withdrawals were largely redeployed into government bills and bonds; not all withdrawals required offsetting, but they contributed to reallocation from weaker to stronger banks and required a stronger policy response by the Fed and Eurosystem.

### VI. Reserve management and financial stability externalities
- Procyclical reserve manager behavior compounded banking sector funding problems when other funding sources dried up.
- Main adverse channels:
  - Sharp reductions in deposits coincided with spikes in 3-month LIBOR – OIS spreads, notably after Lehman Brothers’ collapse.
  - Sales of bank securities by reserve managers (certificates of deposit, commercial paper, covered bonds, equities) added supply pressure, depressing prices and raising funding rates (volumes smaller than deposit moves).
  - Signaling effects: reserve manager withdrawals may signal lower risk tolerance or concerns about counterparties; RMT 2009: ~half of respondents (19 of 37) saw potential for policy conflict within central banks between monetary policy liquidity provision and reserve management risk reduction.
- Strategic asset-allocation implications:
  - Credit is inherently procyclical and should have a less prominent allocation in central bank reserves.
  - Countercyclical assets should be more prominent even if they yield lower returns in normal times.
  - A 3–5 year duration strategy in high-quality liquid government bonds proposed because:
    - (1) Selling longer-dated government bonds in a systemic crisis acts countercyclically (flight to quality).
    - (2) Likely to get good prices in crisis, offsetting some yield give-up vs. credit.
    - (3) Reduces pressure to sell under stress.

### VII. Conclusions and policy implications
- Prior to the crisis, reserve portfolios had become riskier following diversification out of government debt; during the crisis many central banks acted procyclically by reducing exposures to the banking sector and agencies.
- Key aggregate estimates and findings:
  - At least US$500 billion of deposits and other investments pulled out of the banking sector by reserve managers.
  - Central banks (notably Fed and Eurosystem) provided unprecedented liquidity support peaking in the months after Lehman Brothers’ collapse, with at least US$1.8 trillion injected in the four months following Lehman.
  - Net dollar swap lines peaked at nearly US$600 billion at end-2008.
- Policy implications:
  - Greater international coordination between central banks may be needed to manage cross-border stability externalities.
  - Reserve portfolios should prioritize countercyclical assets over procyclical credit exposures.
  - Align reserve managers’ incentives with financial stability objectives and consider governance changes to reduce procyclical behavior.
- Open questions for follow-up:
  - How reserve management responds in the sovereign phase of the crisis.
  - Whether rumors of selling euros or euro-denominated debt constitute procyclical behavior.
  - Data limitations on securities holdings merit further study.

*Source: _wp10150 - References ..............................................................................................................................25*

### References .............................................................................................................

### _wp10150 - References ..............................................................................................................................25 

### I. INTRODUCTION
- Central banks provide funding to the banking sector mainly through two channels: (1) regular monetary refinancing operations (domestic liquidity) and (2) investing official reserves in short-term bank deposits and securities of foreign banks (cross-border funding).
- Prior to the global financial crisis in 2007, a large share of central bank reserves was held in deposits with large international banks; these were significant in absolute terms and concentrated in the largest banks.
- When the crisis hit, many central banks withdrew these investments abruptly, behaving procyclically and sometimes exacerbating banking sector funding problems (see McGuire and von Peter, 2009).
- Paper objective: estimate amounts withdrawn by reserve managers from the private sector, compare with central bank support measures, investigate explanations and implications of procyclical reserve management, and discuss implications for strategic asset allocation.

### II. RESERVE MANAGEMENT OBJECTIVES
- Reserves held primarily for “self-insurance” against financial shocks and enhancing credibility of monetary policy; typically invested in liquid, short-duration government bills and bonds in reserve currencies (U.S. dollar, euro).
- IMF’s Guidelines for Foreign Exchange Reserve Management (2001) state reserve management should ensure:
  - (1) adequate foreign exchange reserves for a defined range of objectives;
  - (2) liquidity, market, and credit risks are controlled prudently;
  - (3) subject to liquidity and other risk constraints, reasonable earnings generated over the medium to long term.
- “Reserve management function” defined as functional areas inside the central bank responsible for composition and risk management of foreign reserves (front and middle office); ownership structures (e.g., ministry of finance) can complicate linkages.

### III. PRE-CRISIS TRENDS IN RESERVE MANAGEMENT
- Major trend: increased focus on return and shift out of government bonds beginning around the turn of the century; diversification into agency securities, deposits, and sometimes equities.
- Global reserves (excluding gold) rose from US$1.8 trillion in 1998 (year-end) to US$8.5 trillion 11 years later.
- Reserve adequacy ratios increased; commonly used indicators: short-term debt up to 12 months and 3 months worth of imports (Greenspan-Guidotti rule and import cover).
- Share of U.S. dollar in global reserves remained around 60 percent over 2003–09; euro around 30 percent (COFER sample ~60 percent of global reserves).
- Drivers of diversification:
  - Rapid reserve accumulation and surplus reserves increasingly managed as national wealth.
  - Low interest rates and global search for yield.
  - Perception (circa late 1990s) of smaller supply of short-term government debt in reserve countries; official reserves sometimes exceeded supply of short-term government debt in main reserve countries.
  - Professionalization of reserve management and benchmarking/peer pressure (e.g., NCBs managing ECB reserves ranked by performance).
- Investment beliefs and risks:
  - Assumption that investment grade credit spreads adequately compensated additional risk and that central banks could hold to maturity.
  - Mean-variance approaches and some investment-bank literature suggested negative marginal risk contribution of investment grade credit, leading to greater credit exposure.
  - Creation of sovereign wealth funds (SWFs) paralleled search for yield; SWFs typically less constrained by liquidity needs than central banks.

### IV. RESERVE MANAGEMENT DURING THE CRISIS
A. Survey Results and Anecdotal Evidence
- RMT 2009 survey results:
  - 76 percent of responding central banks reduced credit limits in response to the crisis.
  - 85 percent reduced maximum maturity of deposits.
  - Many reduced securities lending; summary: “Central banks joined the flight to quality, cutting deposit holdings, tightening credit limits for assets, reducing exposure to counterparties and emphasizing traditional assets, notably top-rated government bonds and gold.”
- Motivations: primarily fear of losses rather than immediate need to use reserves; most central banks did not significantly reduce overall reserves.
- Many central banks favored holding to maturity (semi-active) rather than active selling; losses on non-government assets often offset by large positive returns on government bond portfolios.
- RMT 2009 found “little evidence of future appetite for diversification into new currencies or asset classes.”

B. Magnitude of the Risk Reduction
- Key data sources: IMF SDDS reserves template (63 countries plus ECB/Eurosystem reporting at end-2009, roughly half of world reserves), TIC, BIS International Banking Statistics, central bank websites.
- SDDS breakdown allows securities vs. deposits; deposits split between (i) other NCBs/BIS/IMF and (ii) commercial banks.
- Bank deposit dynamics (SDDS subscribers):
  - Trend: increase in deposits at commercial banks pre-crisis driven by growing reserves; percentage share flat in 18 months before crisis.
  - Around August 2007 trend changed; after Lehman Brothers’ collapse (September 2008) the withdrawal speed increased.
  - In the two months following Lehman’s collapse an amount equivalent to almost US$170 billion was pulled out of deposits.
  - Between July 2007 and March 2009 the share of commercial bank deposits in reserves of SDDS subscribers dropped 12 percentage points, corresponding to an absolute amount of over US$350 billion.
- Distribution across central banks:
  - Only 15 percent of central banks that held deposits did not reduce allocation meaningfully.
  - 43 percent reduced share of bank deposits by more than 20 percentage points from peak to trough between July 2007 and March 2009.
  - Over two-thirds saw declines of more than 10 percentage points.
  - Nearly 80 percent of the sample reduced their deposits by more than half (measured by min/max approach with peak preceding trough).
- Change in deposits vs. change in reserves:
  - Many central banks reduced deposits far more than their change in reserves; several saw reserves increase while pulling out of deposits—indicating changes driven by risk assessment/tolerance rather than spending reserves.
- Repurchase Agreements (reverse repos):
  - From December 2007 some decline in deposits was offset by an increase in net repos; this offset continued and more than offset deposit decline until Lehman, when net repos fell sharply (US$124 billion from end-September to end-December 2008).
  - Over July 2007 – March 2009 net repos declined by over US$40 billion in the dataset.
- Overall reduction in banking sector exposure:
  - BIS International Banking Statistics: cross-border liabilities of reporting banks to “official monetary authorities” fell by over US$420 billion from June 2007 to March 2009 (peaked US$1,442 billion in Dec 2007, fell to US$935 billion in Mar 2009).
  - SDDS shows slight increase (US$68 billion between July 2007 and March 2009) in deposits with other central banks and BIS; BIS decreased exposure to banking sector by SDR 92 billion (~over US$140 billion) between March 2008 and March 2009.
  - Conservative aggregate estimate: pullback by central banks and BIS of more than the equivalent of US$500 billion during the crisis, majority after Lehman Brothers’ collapse; most redeployed into government bills and bonds, smaller share deposited with reserve-issuing central banks.
- Agencies:
  - TIC data: official holdings of U.S. agency debt increased more than tenfold from March 2000 to June 2008; peaked at US$966 billion in June 2008.
  - Between July 2008 and March 2009 official sector reduced holdings of U.S. agency securities by over US$93 billion, during the period of U.S. agency support measures and conservatorship of Fannie Mae and Freddie Mac.
- Procyclicality in credit limit frameworks:
  - Many frameworks rely on external ratings; limits inversely related to creditworthiness (ratings). These frameworks have inherent procyclical elements.
  - Back-of-the-envelope estimate: between peak end-2007 and end-2008, changes in aggregate credit limits may explain a reduction of bank exposure of around 10 percent (based on ECB framework applied conservatively to a sample of ten largest banks).
  - Given actual pullback around 50 percent, much of the reduction driven by active and semi-active risk reduction rather than automatic limit triggers.

### V. LIQUIDITY SUPPORT FOR THE BANKING SECTOR
- Central banks increased volume and scope of lending: longer maturities, foreign currencies, relaxed collateral; asset purchases (e.g., Fed purchases of agency debt, Eurosystem purchases of covered bonds).
- Liquidity injections approximated by monthly changes in central bank claims on domestic banks (IFS database, 176 central banks):
  - Modest measures until late 2007; initial spike after ECB year-end facilities (withdrawn in Jan 2008); larger spikes after Bear Stearns near-collapse; largest surge in the two months following Lehman Brothers’ collapse.
  - Concentration: about 93 percent of first month’s increase in claims after Lehman was supplied by Eurosystem central banks and the Fed.
- Currency dimension: large proportion of liquidity was in foreign currency (mainly U.S. dollar) via swap lines; net dollar amount outstanding through dollar swaps peaked at nearly US$600 billion at end-2008 (Goldberg, Kennedy, and Miu, 2010).
- Scale of injections:
  - Estimated at least US$1.8 trillion of additional liquidity injected into the banking sector by central banks (mainly Fed and Eurosystem) over the four months following Lehman Brothers’ collapse—excluding later direct purchases of private and government securities.
- Interaction with reserve withdrawals:
  - Some of the liquidity support was needed to offset money withdrawn by other central banks (at least US$500 billion).
  - Not all withdrawn funds needed offsetting (most were reinvested in treasuries and stayed in system), but withdrawal contributed to reallocation from weaker to stronger banks and required stronger policy response by Fed and Eurosystem.

### VI. RESERVE MANAGEMENT AND FINANCIAL STABILITY
- Reserve managers’ procyclical withdrawals compounded banking sector funding problems when other funding sources dried up (McGuire and von Peter, 2009).
- Main channels of adverse impact:
  - Sharp and abrupt reduction in reserves held as bank deposits; timing of deposit reductions generally coincided with spikes in 3-month LIBOR – OIS spread (indicator of interbank funding stress)—notably after Lehman Brothers’ collapse (Figure 11).
  - Sale of bank securities by central banks (certificates of deposit, commercial paper, covered bonds, equities) added supply pressure, depressing prices and raising funding rates (though volumes smaller).
  - Signaling effects: central bank actions may signal lower risk tolerance or concerns about counterparties, potentially triggering market responses; RMT 2009 survey: ~half of respondents (19 of 37) saw potential for policy conflict within central banks between monetary policy liquidity provision and reserve management risk reduction.
- Governance and incentive issues:
  - Reserve management often delegated to lower levels with frameworks imposing upper bounds but typically no lower bounds; decisions may be driven by reserve managers seeking to avoid losses or by senior management seeking to avoid reputational damage.
  - Remedies: realign incentives, introduce passive reserve management for certain asset classes, or pursue higher-level decision processes; international coordination may be needed when decisions of one central bank affect others.
- Strategic asset allocation implications:
  - Credit is inherently procyclical—should have a less prominent allocation in central bank reserves; countercyclical assets deserve a more prominent role even if lower return in normal times.
  - A 3–5 year duration strategy in high-quality liquid government bonds is proposed as an attractive alternative to long-credit strategies: (1) selling longer-dated government bonds in a systemic crisis acts countercyclically (flight to quality), (2) likely to get good prices in crisis, offsetting some yield give-up vs. credit, and (3) reduces pressure to sell under stress.

### VII. CONCLUSIONS
- Prior to the crisis, reserve portfolios had become riskier following diversification out of government debt; during the crisis many central banks acted procyclically by reducing exposures to the banking sector and agencies.
- Estimated at least US$500 billion of deposits and other investments pulled out of the banking sector by reserve managers; simultaneously central banks (notably Fed and Eurosystem) provided unprecedented liquidity support peaking in months after Lehman Brothers’ collapse.
- Reserve managers’ focus on protecting their own assets produced signaling and financial stability externalities, shifting burden of stabilizing the global financial system to reserve-issuing central banks.
- Policy implications:
  - Greater international coordination between central banks may be needed to manage cross-border stability externalities.
  - Reserve portfolios should prioritize countercyclical assets over procyclical credit exposures; align reserve managers’ incentives with financial stability objectives.
- Open questions for follow-up: how reserve management responds in the sovereign phase of the crisis; whether rumors of selling euros or euro-denominated debt constitute procyclical behavior; data limitations on securities holdings merit further study.

*Source: _wp10150 - References ..............................................................................................................................25*

### REFERENCES

### _wp10150 - REFERENCES

### Major themes covered by the references
- Central bank reserve management and trends
- Risk management for central banks and public investors
- FX reserve management frameworks, trends, and challenges
- Liquidity provision, market-making, and crisis management by central banks
- Sovereign Wealth Funds: investment objectives and policy issues
- Credit spreads, credit risk, duration, and portfolio optimization
- Global dollar funding, dollar swap lines, and the international policy response
- Surveys and empirical studies of reserve managers and reserve asset composition

### Recurring publication outlets and formats
- BIS Quarterly Review and BIS Working Papers
- BIS Papers and BIS Annual Report
- Central Banking Publications (RBS Reserve Management Trends series)
- Cambridge University Press (edited volumes on Risk Management for Central Banks and Other Public Investors)
- IMF Survey Magazine and IMF guidelines
- ECB press releases and speeches
- Federal Reserve Bank of New York Staff Report
- Central Banking Journal and Brookings Papers on Economic Activity
- Blogs and commentary (Maverecon Blog; Council on Foreign Relations—Follow the Money Blog)
- Standard & Poor’s RatingsDirect study

*Source: _wp10150 - REFERENCES (source PDF filename).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10150.pdf_