## insea2026007

## Source details

**Canonical URL:** [insea2026007](https://www.imf.org/-/media/files/publications/imf-notes/2026/english/insea2026007.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/imf-notes/2026/english/insea2026007.pdf.md)
- [Structured JSON version](/-/media/files/publications/imf-notes/2026/english/insea2026007.pdf.json)

---

### Overview and context
- Gold’s share in central bank reserves rose from 10 percent of reserves in January 2019 to over 22 percent by August 2025, driven overwhelmingly by valuation effects rather than large-scale accumulation of physical gold.
- Between 2018 and 2025, the market value of central bank gold holdings rose by about $3.2 trillion, from roughly $1.2 trillion to $4.5 trillion—an increase of 268 percent—while the aggregate volume of gold held increased by only about 8.5 percent.
- Nearly two‑thirds of the increase in value occurred at central banks that kept the physical quantity of gold broadly unchanged.
- About one‑third of the increase in aggregate gold holdings (about $1.1 trillion) is attributable to central banks that actively purchased gold, expanding their holdings by about 36 percent in volume terms.
- A small share of global holdings—around 1.5 percent—remains with central banks that reduced their physical gold positions (by roughly 16 percent); valuation effects produced a cumulative increase of about 187 percent in the value of their gold holdings.
- RAMP survey evidence:
  - nearly half of respondents retain gold largely as a legacy asset;
  - a further quarter rely on qualitative diversification considerations;
  - only about one-quarter explicitly incorporate gold into their SAA frameworks;
  - very few engage in short-term or tactical adjustments to gold holdings.

### Gold’s fundamental characteristics as a reserve asset
- Strengths:
  - No credit risk and can act as protection against sanctions (if custody arrangements secure deliverability outside the international financial system).
  - Deep trading liquidity in major bullion markets: London daily average trading volume of about $134 billion.
- Comparators for daily trading volume and turnover:
  - US Treasury market’s trading volume: $1.071 trillion (September 2025).
  - Chinese Government Bond market’s trading volume: $70 billion.
  - Combined EU+UK Government Bond market’s turnover: $110 billion in 2024.
- Limitations:
  - Gold is highly volatile and carries significant market risk.
  - Liquidity is materially lower than headline market value suggests when considered against reserve liquidity needs and custody/geographic constraints.
  - Gold’s convenience-yield, opportunity-cost (real rates), and USD pricing channels produce regime- and shock-dependent behavior.

### Hedging, diversification, and safe-haven properties (empirical findings)
- Definitions used:
  - Hedge: uncorrelated or negatively correlated on average.
  - Diversifier: positively (but not perfectly) correlated on average.
  - Safe haven: uncorrelated or negatively correlated in times of market stress (nonpositive correlation in extreme conditions).
- Interest rate risk:
  - Gold is a good hedge against interest rate risk; gold’s negative correlation with the 10-year US yield is the most stable and consistent finding across periods.
  - Gold’s negative correlation with the 10-year US term premium reinforces its flight-to-safety role.
- Equity risk:
  - Gold’s correlation with the S&P 500 was nearly zero from 1998 to 2026 and negative before the COVID-19 pandemic; correlation turned positive since the COVID-19 pandemic.
- Inflation risk:
  - Gold’s correlation with US inflation surprises is predominantly negative except for the post-GFC to pre-pandemic period; role as an inflation hedge is inconsistent and depends on whether inflation surprises coincide with rising real rates.
- Other assets and stress indicators:
  - Correlations with oil, bitcoin, and systemic stress (US CISS) vary across regimes—time-varying and shock-dependent.
- Volatility and geopolitical risk:
  - Gold is not a reliable hedge against equity-market volatility (VIX) or geopolitical risk; patterns are regime-dependent and often inconsistent.
- Tail risk:
  - Kernel density estimates conditional on extreme VIX/geopolitical changes show gold returns broadly symmetric and centered about zero in tails; gold does not exhibit reliable safe haven properties in extreme episodes, though it can outperform certain fixed-income assets in some tail scenarios.

### Market risk, returns, and accounting implications
- Returns and volatility:
  - After the COVID-19 pandemic, monthly average annualized returns for gold reached 23.2 percent.
  - The best performing traditional asset cited (Chinese government bonds) averaged 4.5 percent.
  - Gold’s annualized volatility reaches 14 percent, about twice as high as the most volatile prime long-term government bond after the pandemic.
- Risk contribution:
  - Zulaica (2020) example: risk parity in a portfolio of gold and two-year maturity SDR bonds is reached when gold accounts for 10 percent of the portfolio—i.e., 50 percent of portfolio volatility arises from only 10 percent of gold.
- Risk-adjusted performance:
  - Gold’s Sharpe ratio was only best performing in the pre-GFC period; since then its risk-adjusted advantage is not consistent.
- Hedging gold exposure:
  - Derivatives (forwards, futures, options) can hedge gold price risk; protective puts and collar strategies are examples.
  - Market depth for gold derivatives is limited relative to some central banks’ physical holdings, constraining full-scale hedging and potentially imposing high transaction costs or market impact.
- Valuation gains and fiscal/operational implications:
  - Large unrealized valuation gains from price spikes can produce sizable realized profits on sales, potentially leading to large and uneven profit transfers to government and political pressure to monetize gains.
  - RAMP survey: only about one-quarter of respondents report gold revaluation gains and losses in profit and loss; a similar share records them in equity or other comprehensive income; others use historical cost or hybrid approaches. Accounting choices materially affect profit-and-loss volatility and distribution pressures.

### Liquidity considerations and reserve management frameworks
- Tranching approach:
  - Reserves typically split into a liquidity tranche (ammunition for immediate needs) and an investment tranche (focused on returns).
  - Under a tranching framework, gold is ill‑suited for the liquidity tranche and should be confined to the investment tranche.
- Non-tranching approach:
  - Where tranching is not used, incorporate liquidity‑risk metrics directly into asset allocation decisions.
  - Compare liquidity‑adjusted market value of reserve assets with expected liquidity needs; such indicators show gold provides materially less effective liquidity than headline market value suggests.
- Sanctions hedge:
  - Gold can hedge sanction risk if held in custody arrangements that insulate deliverability, but domestic vaulting trades off immediate usability and renders holdings extremely difficult to mobilize rapidly for FX liquidity, international payments, or market interventions.

### Domestic Gold Purchases (DGPs): drivers, strategies, and risks
- Drivers and empirical distribution:
  - Some central banks converted FX assets into gold.
  - Many institutions engaged in DGPs to de‑shadow and support local mining sectors, or respond to elevated international gold prices.
  - Bulk of net purchases in volume terms since 2018 originated in countries without domestic gold production; most gold-producing countries kept holdings broadly unchanged or reduced them.
- Modalities and operational pathways:
  - Monetary gold: already meets international reserve standards and is readily marketable.
  - Nonmonetary gold: requires upgrading through refining, assaying, and certification before qualifying as a reserve asset.
- Risks and governance concerns:
  - Mandate clarity and governance concerns.
  - Balance-sheet exposure and income volatility when non‑monetary gold is acquired and must be refined/monetized before qualifying as monetary gold.
  - Financial integrity and operational arrangements for processing and monetizing domestic gold.
  - Monetary policy implementation complications—DGPs can resemble quasi‑fiscal operations and sit uneasily with central bank mandates.
- Practical implication:
  - Domestic purchases—especially of non‑monetary gold—should generally be avoided or assigned to other public entities better equipped to manage commercial and fiscal risks.

### Legal considerations and recommended safeguards for DGPs
- Core recommendation:
  - Prohibit or phase out central bank engagement in DGPs involving nonmonetary, unrefined gold because they constitute quasi-fiscal activities (QFAs) and risk undermining central bank core mandates.
- If a central bank operates a DGP:
  - adopt a time-bound exit strategy and transfer activity to an appropriate public body (for example, a state gold purchasing agency under the competent ministry);
  - amend central bank legislation to expressly prohibit QFAs, including purchase of domestically produced nonmonetary, unrefined gold for fiscal policy purposes;
  - if purchasing unrefined gold is a public policy priority, assign the function to a public entity other than the central bank.
- Where political constraints force continuation:
  - Ground activity in explicit legal basis with robust safeguards; narrowly define purpose, scope, and limits; frame DGP as exceptional and temporary; require formal coordination with relevant public authorities.
- Financial safeguards to codify:
  - Purchases below prevailing market prices pursuant to explicit pricing rules (with appropriate discounts for assay, logistics, refining, and other costs).
  - Quantitative exposure caps.
  - Timely monetization via prompt refining into monetary gold of at least 995/1,000 purity aligned with international reserve standards, or sale for freely convertible foreign currencies.
  - Ensure subsidies or other financial support to miners are provided outside the central bank.
  - Define loss-allocation arrangements through the central bank’s profit distribution rules and recapitalization mechanisms.
- Transparency and accountability:
  - Mandate regular reporting to Parliament and Ministry of Finance on volumes purchased, realized/unrealized financial outcomes, risk exposures, and risk mitigation measures.
  - Require public disclosure via the central bank’s annual report, dedicated publications, and website.
  - Require independent external audits of the DGP program and periodic internal audits focused on governance, risk management, and internal controls.

### Operational, financial integrity, and monetary policy implications of DGPs
- Financial and valuation risks:
  - DGPs can generate significant balance sheet risks if pricing, valuation, and risk-control frameworks are not carefully designed.
  - Nonmonetary gold entails additional costs: assay, refining, transportation, storage, and insurance; prolonged holding periods before upgrading increase illiquidity and valuation uncertainty.
  - Purchasing monetary gold domestically reduces—but does not eliminate—these risks because the asset is immediately tradable and can be valued using international price benchmarks.
- Financial integrity (AML/CFT) risks:
  - Gold from artisanal or small-scale mining or through complex intermediary chains can be used to launder illicit proceeds or circumvent sanctions and tax obligations.
  - Central banks purchasing directly assume responsibility for customer due diligence, beneficial ownership verification, transaction monitoring, and recordkeeping.
  - Risks are overwhelmingly concentrated in programs involving nonmonetary gold.
- Operational and logistical risks:
  - Sourcing, assaying, refining, transportation, custody, and insurance require specialized capabilities many central banks do not possess.
  - Gold failing to meet London Good Delivery standards may face liquidity constraints, price discounts, or exclusion from key markets.
- Monetary policy and liquidity management:
  - Purchasing gold from domestic counterparties injects local currency into the economy, expanding the monetary base and typically requiring sterilization.
  - Typical sterilization instrument: a liquidity-absorbing open market operation at the policy rate.
  - Where sterilization capacity is limited, sterilization may be costly and conducted through FX sales, which increase the proportion of gold in reserves and affect FX market conditions.
  - Another option: sell domestically acquired gold for FX and use the proceeds for sterilization, effectively acting as a gold trader rather than a traditional reserve manager.

### Practical guidance and policy recommendations
- Treat gold as a high‑risk reserve asset.
- Avoid interpreting valuation gains from higher gold prices as a permanent increase in reserves or as a substitute for liquid foreign assets.
- Use tranching: confine gold to the investment tranche; do not include gold in the liquidity tranche.
- Where tranching is absent, explicitly integrate liquidity‑adjusted market values and expected liquidity needs into allocation decisions to reflect gold’s lower effective liquidity.
- Limit the share of gold in reserve portfolios and consider small allocations (typically below 5 percent) for volatility/tail‑risk mitigation in low-duration portfolios; higher allocations may be justified for longer-duration portfolios or those exposed to non-reserve currencies—but recognize regime dependence.
- When considering hedging, weigh the limited market depth and potential high costs of gold derivatives; plan hedging programs that acknowledge partial feasibility and market-impact constraints.
- Avoid domestic gold purchase programs within central banks—especially purchases of non‑monetary gold—or assign such activities to other public entities that can better manage commercial, fiscal, and operational risks.
- Ensure robust governance, mandate clarity, and accounting transparency to manage valuation swings, profit transfer frameworks, and political pressures related to gold revaluations.

### Box 1 — Optimal share of gold in portfolios (key results)
- Method: Markowitz optimization combining gold with government bonds of traditional reserve currencies over May 2006–October 2025.
- Short-duration portfolio (gold + three-month bonds):
  - Minimal volatility obtained with zero percent of gold.
  - Maximum return reached with 31 percent of gold (at substantially higher volatility).
- Long-duration portfolio (gold + 7–10-year bonds):
  - Minimum volatility reached with 34 percent of gold (at the expense of lower return).
  - Maximum return reached with zero percent of gold.
- Practical implication:
  - Optimal level of gold likely lies between zero and one-third, depending on portfolio duration and unit of account.

### Box 2 — Haircut on gold (VaR methodology and findings)
- VaR definition:
  - VaR(Δ Gold returns,훼,Δ푇) = −(휇Δ푇 + 휙−1(1−훼) 휎√Δ푇), with parameters defined in the note.
- Implementation benchmarks:
  - Basel III-style reference: US implementation calibrates haircuts at VaR 99 percent with a 10-business-day holding period; for gold (grouped with main index equities), prescribed haircut is 15.0 percent; for US Treasury > five years, prescribed haircut is 4.0 percent.
- Parameter choices used:
  - 훼 = 99% and Δ푇 = 2 days.
- Empirical haircut estimates (regime dependent):
  - In stressed markets (GFC), haircut ≈ 12 percent using stressed volatility (conservative).
  - In stressed markets, haircut ≈ 9.5 percent based on historical average volatility.
  - In relatively calm markets (post-pandemic), haircut ≈ 8 percent in the conservative scenario.
  - In relatively calm markets, haircut ≈ 5 percent otherwise.
  - These haircuts are about twice as high as the haircut for a long-duration 10-year Treasury note using the same methodology.
- Policy recommendation:
  - Recommend applying a haircut of 6–12 percent to the market value of gold included in the liquidity tranche, based on VaR 99 percent with stressed volatility over 1998–2025.
- Storage cost example:
  - Bank of England charges GBP 0.035 per bar per night, or GBP 12.775 per bar per year; that is, $17.245 per bar per year in September 2025, or about $0.043 per Troy ounce (a gold bar is roughly 400 ounces).
  - Average price of gold $3,658.09 per Troy ounce in September 2025 (morning (AM) quoting); yearly storage cost about 0.000012 percent of the value (negligible).
  - Caveat: cost of converting mined gold to monetary gold can be substantial and is not accounted for here.

### Box 3–5 (operational and good-practice initiatives — highlights)
- Box 3 (strategies): three key attributes of DGP strategies are source of gold, process of upgrading to monetary gold, and ultimate use; sources include unrefined artisanal production, jewelry/scrap, and refined domestic output.
- Box 4 (legal safeguards): recommend prohibition/phase-out of central bank DGPs involving nonmonetary gold; if unavoidable, require legal basis, narrow purpose, quantitative caps, prompt upgrading to 995/1,000 purity, reporting, and independent audits.
- Box 5 (London Principles and limitations):
  - London Principles aim to formalize artisanal mining, endorse refining at LBMA Good Delivery refineries, and support environmental, social, and governance standards.
  - Limitations: Principles are mostly silent on central bank-specific risks (financial, liquidity, monetary policy, quasi-fiscal activities).
  - Key policy implication: purchasing monetary gold is always preferable to purchasing nonmonetary gold from a reserve-management perspective; DGPs should generally be approached with caution.

*Source: insea2026007 - Conclusion (pages referenced as provided in the source content).*

### Conclusion .............................................................................................................

### Conclusion

### Document structure and location
- Conclusion appears on page "25".
- References appear on page "26".

### Boxes included in the document
- 1. What Is the Optimal Share of Gold in a Portfolio of Foreign Reserves? .......................................... 17
- 2. Haircut on Gold ................................................................................................................................ 19
- 3. Typical Central Bank Strategies in Domestic Gold Purchases ........................................................ 21
- 4. Legal Considerations to Address Domestic Gold Purchases by Central Banks .............................. 22
- 5. Initiatives to Develop Good Practices of Central Banks’ Domestic Gold Purchases ....................... 23

### Figures included in the document
- 1. Gold Price ........................................................................................................................................... 6
- 2. Changes in Central Banks’ Gold Reserves ........................................................................................ 7
- 3. Daily Average Gold Market Trading Volume in London ..................................................................... 8
- 4. What Is Gold Correlated With? ........................................................................................................ 11
- 5. Gold, Market Volatility, and Geopolitical Risk .................................................................................. 12
- 6. Gold and Tail Risk ............................................................................................................................ 13
- 7. Asset Performance ........................................................................................................................... 14

*Source: insea2026007 - Conclusion (pages referenced as provided in the source content).*

### 8. Gold for Portfolio Diversification ..................................................................................

### 8. Gold for Portfolio Diversification

### Overview and context
- Gold’s share in central bank reserves rose from 10 percent of reserves in January 2019 to over 22 percent by August 2025, driven overwhelmingly by valuation effects rather than large-scale accumulation of physical gold.
- Between 2018 and 2025, the market value of central bank gold holdings rose by about $3.2 trillion, from roughly $1.2 trillion to $4.5 trillion—an increase of 268 percent—while the aggregate volume of gold held increased by only about 8.5 percent.
- Nearly two‑thirds of the increase in value occurred at central banks that kept the physical quantity of gold broadly unchanged, implying that the rise largely reflected revaluation gains from higher gold prices.
- About one‑third of the increase in aggregate gold holdings (about $1.1 trillion) is attributable to central banks that actively purchased gold, expanding their holdings by about 36 percent in volume terms.
- A small share of global holdings—around 1.5 percent—remains with central banks that reduced their physical gold positions (by roughly 16 percent), yet even for these institutions higher gold prices more than offset volume declines, resulting in a cumulative increase of about 187 percent in the value of their gold holdings.
- RAMP survey evidence: nearly half of respondents retain gold largely as a legacy asset; a further quarter rely on qualitative diversification considerations; only about one-quarter explicitly incorporate gold into their SAA frameworks; very few engage in short-term or tactical adjustments to gold holdings.

### Gold’s fundamental characteristics as a reserve asset
- Strengths:
  - No credit risk and can act as protection against sanctions (if custody arrangements secure deliverability outside the international financial system).
  - Deep trading liquidity in major bullion markets: London daily average trading volume of about $134 billion.
- Comparators for daily trading volume and turnover:
  - US Treasury market’s trading volume: $1.071 trillion (September 2025).
  - Chinese Government Bond market’s trading volume: $70 billion.
  - Combined EU+UK Government Bond market’s turnover: $110 billion in 2024.
- Limitations:
  - Gold is highly volatile and carries significant market risk.
  - Liquidity is materially lower than headline market value suggests when considered against reserve liquidity needs and custody/geographic constraints.
  - Gold’s convenience-yield, opportunity-cost (real rates), and USD pricing channels produce regime- and shock-dependent behavior.

### Hedging, diversification, and safe-haven properties (empirical findings)
- General definitions used:
  - Hedge: uncorrelated or negatively correlated on average.
  - Diversifier: positively (but not perfectly) correlated on average.
  - Safe haven: uncorrelated or negatively correlated in times of market stress (nonpositive correlation in extreme conditions).
- Key empirical patterns:
  - Interest rate risk:
    - Gold is a good hedge against interest rate risk; gold’s negative correlation with the 10-year US yield is the most stable and consistent finding across periods.
    - Gold’s negative correlation with the 10-year US term premium reinforces its flight-to-safety role.
  - Equity risk:
    - Gold’s correlation with the S&P 500 was nearly zero from 1998 to 2026 and negative before the COVID-19 pandemic; correlation turned positive since the COVID-19 pandemic, implying weakened hedging against equity downturns.
    - Post-2022 equity declines driven primarily by monetary tightening and supply-driven inflation surprises—raising real rates and weighing on gold via the opportunity-cost channel.
  - Inflation risk:
    - Gold’s correlation with US inflation surprises is predominantly negative except for the post-GFC to pre-pandemic period; gold’s role as an inflation hedge is inconsistent and depends on whether inflation surprises coincide with rising real rates.
  - Other assets and stress indicators:
    - Correlations with oil, bitcoin, and systemic stress (US CISS) vary across regimes—time-varying and shock-dependent.
  - Volatility and geopolitical risk:
    - Gold is not a reliable hedge against equity-market volatility (VIX) or geopolitical risk; correlation patterns are regime-dependent and often inconsistent.
  - Tail risk:
    - Kernel density estimates conditional on extreme VIX/geopolitical changes show gold returns broadly symmetric and centered about zero in tails; gold does not exhibit reliable safe haven properties in extreme episodes, though it can outperform certain fixed-income assets in some tail scenarios.

### Market risk, returns, and accounting implications
- Returns and volatility:
  - After the COVID-19 pandemic, monthly average annualized returns for gold reached 23.2 percent.
  - The best performing traditional asset cited (Chinese government bonds) averaged 4.5 percent.
  - Gold’s annualized volatility reaches 14 percent, about twice as high as the most volatile prime long-term government bond after the pandemic.
- Risk contribution:
  - Zulaica (2020) example: risk parity in a portfolio of gold and two-year maturity SDR bonds is reached when gold accounts for 10 percent of the portfolio—i.e., 50 percent of portfolio volatility arises from only 10 percent of gold.
- Risk-adjusted performance:
  - Gold’s Sharpe ratio was only best performing (relative to other assets) in the pre-GFC period; since then its risk-adjusted advantage is not consistent.
- Hedging gold exposure:
  - Derivatives (forwards, futures, options) can hedge gold price risk; protective puts and collar strategies are examples.
  - Market depth for gold derivatives is limited relative to some central banks’ physical holdings, constraining full-scale hedging and potentially imposing high transaction costs or market impact.
- Valuation gains and fiscal/operational implications:
  - Large unrealized valuation gains from price spikes can produce sizable realized profits on sales, potentially leading to large and uneven profit transfers to government and political pressure to monetize gains.
  - RAMP survey: only about one-quarter of respondents report gold revaluation gains and losses in profit and loss; a similar share records them in equity or other comprehensive income; others use historical cost or hybrid approaches. Accounting choices materially affect profit-and-loss volatility and distribution pressures but do not eliminate underlying tensions.

### Liquidity considerations and reserve management frameworks
- Tranching approach:
  - Reserves typically split into a liquidity tranche (ammunition for immediate needs) and an investment tranche (focused on returns).
  - Under a tranching framework, gold is ill‑suited for the liquidity tranche and should be confined to the investment tranche.
- Non-tranching approach:
  - Where tranching is not used, incorporate liquidity‑risk metrics directly into asset allocation decisions.
  - Compare liquidity‑adjusted market value of reserve assets with expected liquidity needs; such indicators show gold provides materially less effective liquidity than headline market value suggests.
- Sanctions hedge:
  - Gold can hedge sanction risk if held in custody arrangements that insulate deliverability, but domestic vaulting trades off immediate usability and renders holdings extremely difficult to mobilize rapidly for FX liquidity, international payments, or market interventions.

### Domestic Gold Purchases (DGPs) and related risks
- Drivers of DGPs:
  - Some central banks converted FX assets into gold.
  - Many institutions engaged in DGPs to de‑shadow and support local mining sectors, or respond to elevated international gold prices.
- Empirical distribution:
  - Bulk of net purchases in volume terms since 2018 originated in countries without domestic gold production; most gold-producing countries kept holdings broadly unchanged or reduced them.
- Risks and challenges from DGPs:
  - Mandate clarity and governance concerns.
  - Balance-sheet exposure and income volatility when non‑monetary gold is acquired and must be refined/monetized before qualifying as monetary gold.
  - Financial integrity and operational arrangements for processing and monetizing domestic gold.
  - Monetary policy implementation complications—DGPs can resemble quasi‑fiscal operations and sit uneasily with central bank mandates.
- Practical implication:
  - Domestic purchases—especially of non‑monetary gold—should generally be avoided or assigned to other public entities better equipped to manage commercial and fiscal risks.

### Practical guidance and policy recommendations
- Treat gold as a high‑risk reserve asset.
- Avoid interpreting valuation gains from higher gold prices as a permanent increase in reserves or as a substitute for liquid foreign assets.
- Use tranching: confine gold to the investment tranche; do not include gold in the liquidity tranche.
- Where tranching is absent, explicitly integrate liquidity‑adjusted market values and expected liquidity needs into allocation decisions to reflect gold’s lower effective liquidity.
- Limit the share of gold in reserve portfolios and consider small allocations (typically below 5 percent) for volatility/tail‑risk mitigation in low-duration portfolios; higher allocations may be justified for longer-duration portfolios or those exposed to non-reserve currencies—but recognize regime dependence.
- When considering hedging, weigh the limited market depth and potential high costs of gold derivatives; plan hedging programs that acknowledge partial feasibility and market-impact constraints.
- Avoid domestic gold purchase programs within central banks—especially purchases of non‑monetary gold—or assign such activities to other public entities that can better manage commercial, fiscal, and operational risks.
- Ensure robust governance, mandate clarity, and accounting transparency to manage valuation swings, profit transfer frameworks, and political pressures related to gold revaluations.

*IMF staff: “Gold in Central Bank Reserves: Strategic Considerations, Market Risks, and Practical Guidance,” Istvan Mak and Etienne Vaccaro‑Grange, July 2026.*

### Box 1. What Is the Optimal Share of Gold in a Portfolio of Foreign Reserves?

### Box 1. What Is the Optimal Share of Gold in a Portfolio of Foreign Reserves?

### Optimal share by portfolio duration
- Method: Markowitz portfolio optimization on portfolios combining gold with government bonds of traditional reserve currencies (US, German, Swiss, Japanese, and Chinese bonds). Two portfolios constructed: short-duration (gold + three-month bonds) and long-duration (gold + 7–10-year bonds). Share of gold varied from 0 to 100 percent. Returns and volatility assessed using end-of-the-month observations from May 2006 to October 2025. 7–10-year bond returns from ICE BofA 7–10-year total return index; three-month returns derived from zero-coupon yield curves.
- Short-duration portfolio (liquidity tranche):
  - Minimal volatility obtained with zero percent of gold because gold is much more volatile than three-month bonds.
  - Zero percent gold yields the smallest return of all portfolio combinations.
  - Maximum return is reached with 31 percent of gold, but this comes with substantially higher portfolio volatility.
  - Conclusion: gold materially increases volatility for short-duration allocations; volatility of the short-duration portfolio is nearly entirely driven by the share of gold.
- Long-duration portfolio (investment tranche):
  - Minimum volatility reached with 34 percent of gold in the example using 7–10-year bonds.
  - The lower volatility at 34 percent gold comes at the expense of a lower return.
  - Maximum return for the long-duration portfolio is reached with zero percent of gold because gold did not beat long-term bond performance over 2006–25.
- Practical implication:
  - An optimal gold share for short-duration portfolios to maximize return is 31 percent (at the cost of higher volatility).
  - An optimal gold share for long-duration portfolios to minimize volatility is 34 percent (at the expense of return).
  - In practice, reserve portfolios combine short, medium, and long durations; the optimal level of gold likely lies between zero and one-third, depending on portfolio duration and unit of account (numeraire).
  - Zulaica (2020) evidence: gold’s diversification benefits are enhanced in the presence of foreign exchange risk (optimal share depends on the portfolio’s unit of account).

*Source: IMF staff calculations.*

### Key takeaway
- Gold can improve portfolio resilience when combined with other reserve assets, but the optimal share depends critically on portfolio duration and risk preferences.  

---

### Operational drawback: gold and the liquidity front line
- Liquidity frameworks for reserves:
  - Tranching: separates reserves into a liquidity tranche (highly liquid, short-duration, low-risk assets) and an investment tranche (longer-term, potentially more volatile assets).
  - Liquidity-risk metrics: apply risk-based haircuts to market values to derive liquidity-adjusted market values when tranching is not used.
- Role of gold under tranching:
  - Gold’s high price volatility and potential for valuation losses during market stress make it ill-suited for the liquidity tranche; it should be confined to the investment tranche.
- Role of gold under liquidity-risk metrics:
  - Apply risk-based haircuts to gold market values to reflect uncertainty about liquidity over short horizons. Box 2 provides a value-at-risk–based methodology to estimate such haircuts and derive a liquidity-adjusted market value, ensuring insurance equivalence with traditional liquid fixed-income assets and preventing overstatement of effective reserve liquidity.

---

### Box 2 — Haircut on gold (VaR methodology and findings)
- VaR definition for gold returns (as presented):
  - VaR(Δ Gold returns,훼,Δ푇) = −(휇Δ푇 + 휙−1(1−훼) 휎√Δ푇), where 휇 is the average return on gold prices, Δ푇 is the time required to fully liquidate the asset, 휎 is the volatility of gold returns, 훼 is a confidence level, and 휙 is the cumulative density distribution of the standard normal distribution.
  - Volatility 휎 can be historical or estimated with GARCH models; GARCH can capture volatility clustering and asymmetric responses to market shocks. Using volatility at a confidence interval (for example, same as VaR) captures stressed market conditions.
- Implementation notes and benchmarks:
  - Haircut defined as VaR at chosen 훼 and Δ푇: Haircut = VaR(Δ Gold returns,훼,Δ푇).
  - GARCH haircuts combine a stressed volatility estimate—the 99th percentile of the GARCH conditional volatility series—with a 99th percentile Student-t quantile.
  - Period definitions for regime analysis: pre-GFC (January 1998–July 2007), GFC (August 2007–June 2009), post-GFC to pre-pandemic (July 2009–February 2020), pandemic (February 2020–December 2021), post-pandemic (January 2022–January 2026). Data are based on daily observations.
  - Comparable Basel III-style reference: US implementation calibrates haircuts at VaR 99 percent with a 10-business-day holding period and historical volatility. For US Treasury securities residual maturity > five years, prescribed haircut is 4.0 percent. For gold (grouped with main index equities), prescribed haircut is 15.0 percent.
- Parameter choices and rationale used:
  - 훼 = 99% and Δ푇 = 2 days (interest in potential loss at 99 percent confidence over a two-day liquidation horizon). Two-day liquidation is reasonable if gold is of monetary standard and held at the Bank of England or at the London Bullion Market Association; longer Δ푇 if gold is held domestically in central bank vaults.
- Empirical haircut estimates (regime dependent):
  - In stressed markets (for example, during the global financial crisis), haircut ≈ 12 percent using stressed volatility (conservative scenario).
  - In stressed markets, haircut ≈ 9.5 percent based on historical average volatility.
  - In relatively calm markets (for example, after COVID-19 pandemic), haircut ≈ 8 percent in the conservative scenario.
  - In relatively calm markets, haircut ≈ 5 percent otherwise.
  - Most importantly, these haircuts are about twice as high as the haircut for a long-duration 10-year Treasury note using the same methodology.
- Policy recommendation:
  - As haircuts are insurance against stressed market conditions, recommend applying a haircut of 6–12 percent to the market value of gold included in the liquidity tranche, based on a VaR at 99 percent with stressed volatility run over the period 1998–2025.
- Storage cost note:
  - Central banks storing gold at the Bank of England are charged GBP 0.035 per bar per night, or GBP 12.775 per bar per year; that is, $17.245 per bar per year in September 2025, or about $0.043 per Troy ounce (a gold bar is roughly 400 ounces).
  - With an average price of gold $3,658.09 per Troy ounce in September 2025 (morning (AM) quoting), the yearly storage cost was about 0.000012 percent of the value, which is negligible.
  - Caveat: the cost of converting mined gold to monetary gold (mining, refining, assaying, transportation to the vault) can be substantial and is not accounted for here.

*Source: IMF staff calculations.*

---

### Domestic gold purchases (DGPs): channels, strategies, and risks
- Distinction of modalities:
  - Monetary gold: already meets international reserve standards and is readily marketable.
  - Nonmonetary gold: requires upgrading through refining, assaying, and certification before qualifying as a reserve asset.
  - The modality fundamentally affects mandate, balance sheet, operational capacity, financial integrity, and monetary policy implications.
- Typical central bank strategies in domestic gold purchases (Box 3):
  - Three key attributes define strategies: source of gold, process of upgrading to monetary gold, and ultimate use of acquired gold.
  - Sources: unrefined gold from artisanal or industrial producers; jewelry or scrap from the general public; refined gold from domestic refiners.
  - Upgrading: swap with a refinery for monetary gold; refine domestically if capacity exists; send abroad for processing.
  - Management after acquisition: store in international trading centers (London, New York); hold domestically to enhance reserve visibility; sell portions regularly to maintain stable share of gold in reserves.
- Risks and governance concerns:
  - Procuring nonmonetary (unrefined or semi-processed) gold entails activities resembling commodity trading (quality verification, logistics, pricing, refining decisions).
  - Such activities are operationally and legally distinct from reserve management and resemble commercial or quasi-fiscal functions, raising mandate, balance sheet, accountability, and political pressure risks.
  - Initiatives like the London Principles address environmental, social, and governance practices in artisanal mining but do not resolve core central bank governance issues (mandate consistency, balance sheet protection, accountability for losses, insulation from political pressures).
  - Domestic purchase of monetary gold poses fewer mandate ambiguities but can still raise governance concerns if purchases serve industrial or redistributive objectives rather than reserve-management needs.
- Overarching message:
  - Mandate and governance risks are highest when central banks purchase nonmonetary gold; risks differ in intensity and interaction depending on the form of gold acquired and operational pathways.

---

### Legal considerations and recommended safeguards for DGPs (Box 4)
- Recommendation: prohibit or phase out central bank engagement in DGPs involving nonmonetary, unrefined gold because they constitute quasi-fiscal activities (QFAs) and risk undermining central bank core mandates.
  - If a central bank operates a DGP, adopt a time-bound exit strategy and transfer activity to an appropriate public body (for example, a state gold purchasing agency under the competent ministry).
  - Amend central bank legislation to expressly prohibit QFAs, including purchase of domestically produced nonmonetary, unrefined gold for fiscal policy purposes.
  - If purchasing unrefined gold is a public policy priority, assign the function to a public entity other than the central bank.
- Where political constraints force continuation of a DGP:
  - Ground activity in explicit legal basis with robust safeguards; narrowly define purpose (for example, ultimate conversion into international reserves as monetary gold), scope, and limits.
  - Frame DGP as exceptional and temporary, subject to statutory review or potential transfer outside the central bank.
  - Require formal coordination mechanisms with relevant public authorities (ministries responsible for mining, environment, labor).
- Financial safeguards to codify in law/regulation:
  - Purchases below prevailing market prices pursuant to explicit pricing rules (with appropriate discounts for assay, logistics, refining, and other costs).
  - Quantitative exposure caps.
  - Timely monetization via prompt refining into monetary gold of at least 995/1,000 purity aligned with international reserve standards (often using accredited offshore refineries), or sale for freely convertible foreign currencies.
  - Ensure subsidies or other financial support to miners are provided outside the central bank (for example, by a relevant public agency).
  - Define loss-allocation arrangements through the central bank’s profit distribution rules and recapitalization mechanisms to compensate the central bank for losses under the DGP program.
- Transparency and accountability:
  - Mandate regular reporting to politically accountable bodies (Parliament, Ministry of Finance) on volumes purchased, realized/unrealized financial outcomes, risk exposures, and risk mitigation measures.
  - Require public disclosure via the central bank’s annual report, dedicated publications, and website.
  - Require independent external audits of the DGP program and periodic internal audits focused on governance, risk management, and internal controls applicable to the QFA.

*This box was prepared by Tomoyuki Kubota (IMF Legal Department).*

---

*Source: IMF staff calculations; IMF staff (legal) analysis.*

### Box 5. Initiatives to Develop Good Practices of Central Banks’ Domestic Gold Purchases

### Box 5. Initiatives to Develop Good Practices of Central Banks’ Domestic Gold Purchases

### London Principles: objectives and scope
- Aim to provide a comprehensive framework for central banks to responsibly purchase domestically mined gold through domestic gold purchase programs (DGPs).
- Emphasize: legal frameworks, transparency, community well-being, mercury reduction, due diligence, progressive improvement, formalization, economic incentives, rule of law, accessibility, access to clean finance, proper refining, and international cooperation.
- Supported by central banks from Colombia, Ecuador, Mongolia, and the Philippines.
- Intended benefits: formalize the artisanal and small-scale gold-mining sector, enhance environmental, social, and governance standards, benefit responsible miners, strengthen gold market integrity, and foster collaboration among regulatory and law enforcement agencies.
- Two elements of particular relevance for central banks:
  - Advocacy that gold purchased be refined at LBMA Good Delivery List refineries to ensure tradability on the most liquid market.
  - Suggestion that offered prices should attract the sale of material, taking into account producers’ additional costs from operating a formalized supply chain.

### Limitations of the Principles for central banks
- Most principles address issues relevant to the gold-mining sector, but are mostly silent on critical issues that matter for central banks.
- The principles do not explicitly address central banks’ risks from DGPs: financial risks, potential interference with liquidity management for monetary policy implementation, and quasi-fiscal activities.

### Financial and valuation risks of DGPs
- DGPs can generate significant balance sheet risks if pricing, valuation, and risk-control frameworks are not carefully designed.
- Magnitude of risks depends critically on the form of gold purchased (monetary gold versus nonmonetary gold).
- Specific risk channels:
  - Gold price volatility exposes central banks to valuation swings.
  - Domestic purchases—particularly of nonmonetary gold—entail additional costs: assay, refining, transportation, storage, and insurance.
  - Valuation uncertainty and potentially prolonged holding periods before upgrading to monetary gold.
  - If purchase prices do not adequately discount these costs and risks, DGP programs can generate persistent losses, effectively creating quasi-fiscal transfers that weaken central bank financial autonomy.
  - Illiquidity during the period between acquisition and upgrading to monetary gold amplifies exposure because balance-sheet risks remain unhedged and difficult to unwind, especially when programs are conducted at scale or over extended periods.
  - Purchasing monetary gold domestically reduces—but does not eliminate—these risks because the asset is immediately tradable and can be valued using international price benchmarks, limiting valuation distortions and reducing the likelihood of implicit subsidies.

### Financial integrity risks (AML/CFT)
- Gold—particularly when sourced from artisanal or small-scale mining or through complex intermediary chains—can be used to launder illicit proceeds or circumvent sanctions and tax obligations.
- When central banks purchase gold directly from domestic suppliers, they assume responsibility for robust AML/CFT controls, including:
  - customer due diligence,
  - beneficial ownership verification,
  - transaction monitoring, and
  - recordkeeping that preserves a verifiable audit trail from source to reserve asset.
- Failure to establish and enforce such controls risks incorporating illicit value into official reserves, with potentially severe reputational, legal, and financial consequences.
- Financial integrity risks are overwhelmingly concentrated in programs involving nonmonetary gold (unrefined gold sourced through artisanal mining or complex intermediary chains presents heightened AML/CFT risks).
- Risk profile is substantially lower when purchasing monetary gold domestically with regulated counterparties trading internationally certified bars, though AML/CFT considerations do not disappear entirely.

### Operational and logistical risks
- Operational arrangements for DGPs introduce risks that differ materially from those associated with conventional reserve assets; risks are most acute when nonmonetary gold is acquired.
- Central banks must manage logistics related to sourcing, assaying, refining, transportation, custody, and insurance—often without in-house expertise or established infrastructure.
- Ensuring nonmonetary gold can be refined to internationally accepted standards, transported securely, insured adequately, and custodied in line with reserve management practices requires specialized capabilities many central banks do not possess.
- Gold failing to meet London Good Delivery standards may face liquidity constraints, price discounts, or outright exclusion from key markets, undermining tradability and effective liquidity as a reserve asset.
- Weak custody arrangements, limited insurance coverage, and inadequate verification mechanisms heighten exposure to theft, fraud, or misreporting.
- Domestically purchasing monetary gold with custody and settlement in established international trading centers preserves market access and liquidity and aligns more closely with standard reserve operations, though some operational risks still remain.

### Implications for monetary policy and liquidity management
- Purchasing gold from domestic counterparties injects local currency into the economy, expanding the monetary base.
- As with any central bank asset purchase—including FX—this effect typically requires sterilization to maintain monetary control.
- Typical sterilization instrument in a scenario where central banks acquire and upgrade gold for reserve addition: a liquidity-absorbing open market operation at the policy rate.
- In the context of DGPs:
  - Central banks could face policy or political pressures to purchase gold at a scale or pace not aligned with liquidity management objectives.
  - Where sterilization capacity is limited, sterilization may be considered costly and may be conducted through FX sales; FX sales generally increase the proportion of gold in otherwise-stable reserves and affect FX market conditions.
  - Another option: sell domestically acquired gold for FX and use the proceeds for sterilization, maintaining total reserve amount and proportion of gold unchanged—effectively making the central bank’s activity similar to that of a gold trader, outside traditional central bank functions.
- Interaction between gold purchase operations and liquidity management requires careful integration into the overall monetary and exchange rate policy framework—particularly where operational or market constraints limit sterilization capacity—to avoid unintended interference with policy objectives.

### Starting point and strategic choices for central banks
- For central banks not currently engaged in DGPs, initiating a program centered on nonmonetary gold represents a discrete escalation of mandate, governance, balance sheet, financial integrity, operational, and monetary policy risks relative to purchasing monetary gold.
- For institutions already purchasing nonmonetary gold, transitioning toward exclusively acquiring monetary gold—through direct market purchases or accelerated upgrade of nonmonetary gold to monetary gold—can materially reduce risk exposure.
- Maintaining or expanding nonmonetary DGPs entrenches quasi-fiscal activities and increases balance sheet vulnerability.
- Central banks already purchasing monetary gold domestically face comparatively narrower challenges, manageable through standard reserve governance and risk-management frameworks.

### Key policy implication and recommendations
- Unambiguous policy implication: DGPs should generally be approached with caution; purchasing monetary gold is always preferable to purchasing nonmonetary gold from the reserve management perspective.
- Rationale:
  - Purchasing monetary gold preserves market liquidity.
  - Purchasing monetary gold limits balance sheet and governance risks.
  - Purchasing monetary gold remains broadly consistent with central banking functions.
  - Purchasing nonmonetary gold fundamentally alters the nature of central bank activity, exposes the institution to significant quasi-fiscal and operational risks, and should, in most circumstances, be avoided or assigned to other public entities.
- Decisions on gold accumulation—including through DGPs—should be firmly anchored in SAA and sound policy analysis, rather than operational convenience or short-term considerations.

### Conclusion (from the note)
- Gold has reemerged as a prominent component of central bank reserves, but its role is fundamentally distinct from liquid FX assets.
- Gold’s attributes:
  - Benefits: lack of credit risk, potential to perform during periods of stress.
  - Constraints: high volatility, limited and contingent diversification properties, weak liquidity characteristics.
- These features constrain gold’s effectiveness as a core reserve asset and underscore the need to assess gold as part of a broader portfolio strategy aligned with the central bank’s policy objectives and risk-bearing capacity.
- DGP programs add complexity: they can blur institutional mandates, create quasi-fiscal exposures, strain operational capacity, and complicate monetary policy implementation—risks most acute when acquiring nonmonetary gold.
- For gold-producing economies, excessive accumulation can weaken the self-insurance role of reserves because adverse gold price shocks can simultaneously affect export revenues and reserve valuations.
- Determining the appropriate share of gold requires a holistic, risk-based assessment of external vulnerabilities, economic structure, and policy frameworks.

*Source: World Gold Council and IMF staff.*

---


_Source: https://www.imf.org/-/media/files/publications/imf-notes/2026/english/insea2026007.pdf_