## ta-handbook-chapter-foreign-exchange-reference-rates

## Source details

**Canonical URL:** [ta-handbook-chapter-foreign-exchange-reference-rates](https://www.imf.org/-/media/files/publications/miscellaneous/english/2025/ta-handbook-chapter-foreign-exchange-reference-rates.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/miscellaneous/english/2025/ta-handbook-chapter-foreign-exchange-reference-rates.pdf.md)
- [Structured JSON version](/-/media/files/publications/miscellaneous/english/2025/ta-handbook-chapter-foreign-exchange-reference-rates.pdf.json)

---

### Executive Summary
- Purpose: provide comprehensive guidance on the design, implementation, and maintenance of robust FX reference rate frameworks for the FX spot market.
- Key objectives: enhance market integrity, support monetary policy transmission, and foster market development.
- Principal challenges identified:
  - data quality and availability
  - market fragmentation
  - methodology selection
  - market manipulation risks
  - technological integration
  - need for ongoing maintenance and adaptation
- Four primary models for FX reference rate determination:
  - central bank-managed fixing sessions;
  - indicative quotes submitted by market makers;
  - quotes (indicative or executable) observed on trading platforms;
  - weighted average exchange rates of FX transactions.
- Design priorities: robustness, representativeness, and replicability.
- Governance: central bank often best placed as administrator; governance should include oversight committees, transparency, regular reviews, and audits.
- Data sufficiency: guidelines and contingency methods when data are insufficient or not timely (executable quotations, cross rates, previous day’s benchmarks, expert judgment).
- Transition guidance: prefer a single-step transition to new benchmarks when possible; gradual transitions may be necessary; emphasize communication and stakeholder consultation.
- Transparency recommendations: publish detailed methodology statements, disclose contingency methodology usage, and provide channels for stakeholder feedback.

### Context and Issue
- Primary purpose: provide the public with representative exchange rates and contribute to market development by improving price discovery and transparency.
- Potential administrators: data providers, trade associations, regulated exchanges, trading platforms, and central banks.
- Typical TA request contexts: FX policy formulation; transition to a new exchange rate regime; market development objectives; implementing international best practice; considerations under Article VIII, Section 3.
- Methodology importance: methodology affects representation of market conditions and volatility; especially relevant in transitions from fixed to floating exchange rates.
- Main implementation challenges:
  - data from wholesale market often not timely nor fully representative; liquidity issues in emerging markets
  - market fragmentation when order flows concentrate among a few institutions
  - choice of methodology (average, weighted, median, quotes, transactions) affects outcomes and must balance accuracy, transparency, and relevance
  - weak institutional frameworks increase manipulation risk
  - technological integration and infrastructure needs
  - need for ongoing maintenance and periodic review
- Policy interaction: central bank FX policy responses highlight the need for transparent calculation methodologies.

### Framework Design — Main Design Models
- Four key models described:
  1. Central bank-managed fixing sessions
     - used in cases of extreme market dysfunction; similar to interbank FX auctions; may involve pre-submission of bids and offers; can produce a market-clearing midpoint rate; now rarely used.
  2. Simple arithmetic average of FX spot quotes submitted by a panel of most active domestic banks
     - quotes are indicative; often trimmed of extreme values; treats contributions equally and may introduce bias; limited recent use.
  3. Simple arithmetic average of FX spot quotes (indicative or executable) observed at a point in time or over a short time window on most active trading platforms
     - suitable where continuous trading exists; critical to observe tradable rates not indicative; can be published during the trading session; widely used in emerging markets and by benchmark providers in advanced economies.
  4. Weighted arithmetic average of FX spot transactions observed over a short time window or the entire trading session on the most active trading platform
     - most sophisticated; suitable where transactions are observable.
- Advantages and drawbacks:
  - Fixing sessions: enable endogenous demand/supply discovery in absent interbank markets; risk that opening rate may not be equilibrium; sometimes no trading occurs—then interbank quotations may be used.
  - Panel quotes: supportive in early sessions for price guidance but representativeness not ensured; trimming used to reduce outliers.
  - Quasi–point-in-time platform observations: lower manipulation risk if market deep and heterogeneous; manipulation risk remains if participants can concertedly influence rates at the observation point.
  - Session-long transaction observation: robust to intraday liquidity fluctuations and manipulation attempts; earliest publication is after session end or next business day, which can reduce representativeness and may conflict with IMF Guidance on Multiple Currency Practices.

### IOSCO Principles (Box 2) — Summary and Market-Stage Implications
- Overview:
  - The IOSCO Principles for Financial Benchmarks (published in July 2013) outline 19 principles grouped into four categories to enhance reliability, transparency, and integrity of benchmarks.
- Governance category (five principles)
  - Address conflicts of interest; Administrators retain primary responsibility; establish oversight functions; ensure transparency and accountability.
- Quality of the Benchmark category (five principles)
  - Benchmarks should be based on prices formed by competitive market forces; addresses data sufficiency and a hierarchy of data inputs; ensures transparency of calculations and dissemination.
- Quality of the Methodology category (five principles)
  - Requires publication of minimum methodology information; procedures for material changes; policies for benchmark transitions/cessations.
- Accountability category (four principles)
  - Complaints procedures; audits and audit trails; cooperation with regulatory authorities.
- Market-stage implications:
  - Nascent Markets:
    - inputs predominantly trades between banks and FX bureaus with their clients; benchmark calculated once per day, ex-post, primarily for accounting; may derive from official auctions or fixing sessions; described as robust, moderately representative, but lacking replicability.
  - Emerging, Developing and Maturing Markets:
    - transactions include interbank and wholesale dealer-client deals; benchmarks expand to portfolio valuation and performance measurement; publish rates at regular intervals during the day; include all transactions within window and apply volume weighting; replicability becomes increasingly critical.
  - Established or Mature Markets:
    - active interbank and interdealer trades and electronic trading channels; benchmark computed over short windows—merging point-in-time data with a narrow window—and include mechanisms to filter out outliers; supports derivative hedging, index construction, portfolio valuation.

### Timing, Sample Size, and Publication Practices (from IOSCO)
- Sample window sizing evolves with market sophistication; as markets deepen, windows can be narrowed.
- Trade-offs between window length and client/dealer costs:
  - Longer windows: reduce client execution costs, spread hedging activity, lower price impact; reduce risk-adjusted profitability for dealers.
  - Shorter windows: increase dealer returns but raise client costs.
  - Differences in weighting function within the window play a comparatively minor role.
- Multiple exchange rates:
  - Number of rates per objective should be limited (e.g., only one closing rate per day).
  - Methodologies should be consistent, comparable, and fully documented.
- One-day lag publications:
  - When data submission delays occur, rate may be computed and published with a one-business-day lag; derived exchange rate should be used solely for accounting purposes and communications must state this limitation.
- Participation and data submission:
  - Central bank should specify data requirements, identify submitting institutions, ensure infrastructure, and govern submissions consistent with IOSCO Principle 14 on submitter code of conduct.
  - Distinguish between indicative quote submission and transaction reporting; transaction reporting requires greater infrastructure.
- Best practice: publish a single mid-rate as official reference rate; buy/sell rates may be disseminated but roles must be defined.

### Eligible Transactions, Exclusions, and Spot Convention
- Eligibility criteria:
  - Transactions must involve the same instrument (FX spot) and ideally be sourced from a single marketplace where prices are formed by competitive forces.
  - Criteria may be relaxed under predefined contingency conditions when wholesale market inactive.
- When wholesale market not active:
  - Expand scope to dealer-to-customer trades; set transaction amount thresholds to exclude very small retail transactions.
- Data quality:
  - Better for interbank transactions; retail transactions often operationally difficult and low volume—justifying exclusion.
  - In countries with strict capital controls, retail prices can deviate substantially from wholesale market.
- Spot convention:
  - Spot (T+2) is most commonly used; some markets use T+0 or T+1.
  - Typically one convention should define eligible transactions; using all three requires recalculation into spot using interest rate differential; higher interest rate differential increases measurement error.
- Exclusions:
  - Transactions between central bank and market participants, and between central bank and government, should be excluded.
  - Spot legs of swap transactions should not be considered eligible transactions.
- Very large transactions:
  - Can have outsized impact on volume-weighted averages; solutions include using a median volume-weighted rate or applying thresholds to exclude very large transactions.

### Central Tendency Estimators
- Importance: choosing the measure of central tendency is critical for transaction-based FX benchmarks.
- Unsuitable estimators (e.g., simple mean with skewed data) can produce biased benchmarks.
- Robust alternatives: weighted average, median, or trimmed mean depending on market depth, transaction-size distribution, and manipulation risk.
- Comparison between weighted median and weighted average across five features:
  - Consistent: Weighted Median — Consistent; Weighted Average — Consistent.
  - Robust to outliers: Weighted Median — Highly robust; Weighted Average — Not robust.
  - Unbiased: Weighted Median — Unbiased (under certain conditions); Weighted Average — Typically unbiased.
  - Distribution assumptions: Weighted Median — Less dependent on assumed distribution; Weighted Average — More dependent on assumed distribution.
  - Efficient: Weighted Median — More efficient in terms of absolute deviation; Weighted Average — More efficient in terms of variance.
- Conclusion: Weighted median is superior across most features; weighted average is optimal when parametric distributional assumptions hold.

### Data Sufficiency Criteria and Contingency Methods
- Data sufficiency derived from requirement that prices be driven by competitive forces in arms-length transactions.
- Calibration:
  - Central bank should define tolerance for insufficient data as a percentage of trading days expected to be insufficient; estimate using at least one year of historical data and re-estimate annually.
- Data insufficiency triggers:
  - Number of transacting banks falls below a set threshold.
  - Daily volume is less than a specified percentage of the average daily volume.
  - High concentration where a few banks account for a substantial share of turnover.
- Rules of thumb:
  - Percentage of trading days with top three banks accounting for more than 80 percent of turnover should be low.
  - Percentage of trading days with zero or near-zero trading volume should be as close to zero as possible.
  - Percentage of trading days with fewer than five active banks is an important metric.
- Alternative data sources hierarchy when interbank market inactive:
  - Dealer-to-customer trades (if not already used).
  - Request for Quote (RFQ), or executable or indicative quotations.
  - Transactions with different currency pairs using pre-determined guidance to recalculate cross rates.
  - Previous day's benchmark as an input.
  - Expert judgment (allowed by IOSCO Principle 8 but may undermine credibility and automation).
- Alternative sources should supplement until sufficient eligible transaction data are available, following a top-down input hierarchy.

### Transition Approaches
- Methodology or benchmark framework changes:
  - Ideally executed in a single step; conceptually preferable and may include retroactive calculation of the new benchmark (retroactive calculation may be prevented by lack of historical data).
  - Many central banks prefer gradual changes to avoid excessive volatility, revaluation effects, and impacts on expectations—especially where managed exchange rate regimes exist.
- Gradual transition approach:
  - Compute weighted average of legacy rate and new rate with drifting weights (legacy rate weight starts at 1, new rate at 0, reversed by end).

### Governance, Roles, and Central Bank Considerations
- Key governance roles (from IOSCO Benchmark Governance Framework):
  - Administrator: manages the overall benchmark process and communications.
  - Administrator Oversight Function: independent committee to establish and review methodology and address conflicts.
  - Calculating Agent: calculates benchmark values and processes data.
  - Publishing Agent: disseminates calculated benchmark values.
  - Audit Function: reviews processes, methodologies, and adherence.
- Central bank as administrator — advantages and internal design:
  - Central bank well positioned: reduces conflicts of interest, enhances FX market functioning, and provides credibility and market insight.
  - Assign responsibilities to a functional unit with FX market knowledge; establish an Oversight Committee (OC) with terms of reference, selection criteria, and conflict declarations.
  - OC composition: include market community representatives; recommended central bank majority with central bank representative as head.
- OC tasks and responsibilities:
  - Periodic review of benchmark definition and methodology.
  - Review proposed amendments and termination procedures.
  - Verify adherence to methodology, consider audit results, and oversee remedial actions.
  - Review cases where expert judgment or contingency conditions were applied.
  - Review and address complaints.
- Separation of duties and operational practice:
  - Daily benchmark determination should separate data collection/validation (central bank) from Calculation and Publishing Agent roles.
  - Good practice: Administrator and Calculation Agent from different business areas and reporting lines.
  - Prefer automated benchmark calculations; contingency methodology requires predefined steps and management approval before dissemination.
- Accountability and transparency:
  - Recommended disclosures on central bank website: methodology statement, contingency triggers, dissemination times, error handling, and contact email for complaints.
  - Audit arrangements: appoint internal or external auditor; consider publishing a statement of compliance with “IOSCO principles for Financial Benchmarks” and appoint external auditor for periodic assurances.
  - Complementary publication: aggregated indicators on market turnover to inform users about representativeness and liquidity.
- Prevention of market manipulation:
  - Risks: fixing orders can create incentives and capabilities for manipulation; point-in-time methodologies more vulnerable if activity is concentrated.
  - Detection: supervisory methods (law enforcement evidence) and observatory methods (analyze trades, quotes, contributions); monitor sensitive dates/times.
  - Prevention strategies: align benchmark design with usage; lengthen observation windows; eliminate motivation to manipulate (e.g., discourage fixing orders).
- Cross rates:
  - Cross rates derive local currency value against non-dominant currencies typically via a common currency (typically the USD).
  - For illiquid cross pairs, central bank rates often indicative; recommended use of mid-rate for valuation and accounting with clear methodology communication.

### Technical Assistance (recommended TA approach)
- Core TA steps:
  - Map out the relevant objectives and constraints.
  - Determine the design parameters of the methodology.
  - Design a data collection process, if needed.
  - Draft a public policy document.
- A. Map Out Objectives and Constraints:
  - Perform quantitative and qualitative evaluation of FX market development; use granular, ideally transaction-level, data.
  - Assess importance of market segments and the parallel market via surveys/anecdotal evidence.
  - Survey established market participants on market functioning indicators.
  - Review legal/regulatory framework and interview central bank subject matter experts and stakeholders.
  - Set objectives and constraints to prioritize representativeness, reliability, and replicability.
- B. Determine Design Parameters:
  - Select reference rate model most robust and representative given market conditions and willingness to rely on contingency methodologies.
  - Back-test historical data to set parameters: eligible transactions, transaction-size thresholds, sample window, central tendency estimator, publication choices, data sufficiency criteria, and contingency prioritization.
  - Objective: balance representativeness, robustness, and replicability rather than maximize stability.
- C. Design Data Collection Process:
  - Use a common reporting template to standardize granular data enabling filtering and cross-checking.
  - Ensure data management includes access control, audit trail, and data retention to distinguish legitimate data cleaning from manipulation.
- D. Produce a Public Policy Document:
  - Document should outline governance and methodology, explain decision-making responsibilities, enable effective audit, and include error handling, communication procedures, and stakeholder consultation arrangements.

*Foreign Exchange Reference Rates — IMF | MCM Technical Assistance Handbook (chapter content provided).*

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

### ta-handbook-chapter-foreign-exchange-reference-rates - References

### Executive Summary
- Purpose: provide comprehensive guidance on the design, implementation, and maintenance of robust FX reference rate frameworks for the FX spot market.
- Key objectives: enhance market integrity, support monetary policy transmission, and foster market development.
- Principal challenges identified:
  - data quality and availability
  - market fragmentation
  - methodology selection
  - market manipulation risks
  - technological integration
  - need for ongoing maintenance and adaptation
- Four primary models for FX reference rate determination: central bank-managed fixing sessions; indicative quotes submitted by market makers; quotes (indicative or executable) observed on trading platforms; weighted average exchange rates of FX transactions.
- Design priorities: robustness, representativeness, and replicability.
- Governance: central bank often best placed as administrator; governance should include oversight committees, transparency, regular reviews, and audits.
- Data sufficiency: guidelines and contingency methods when data are insufficient or not timely (executable quotations, cross rates, previous day’s benchmarks, expert judgment).
- Transition guidance: prefer a single-step transition to new benchmarks when possible; gradual transitions may be necessary; emphasize communication and stakeholder consultation.
- Transparency recommendations: publish detailed methodology statements, disclose contingency methodology usage, and provide channels for stakeholder feedback.

### Context and Issue
- Primary purpose: provide the public with representative exchange rates and contribute to market development by improving price discovery and transparency.
- Potential administrators: data providers, trade associations, regulated exchanges, trading platforms, and central banks.
- Typical TA request contexts: FX policy formulation; transition to a new exchange rate regime; market development objectives; implementing international best practice; considerations under Article VIII, Section 3.
- Methodology importance: methodology affects representation of market conditions and volatility; especially relevant in transitions from fixed to floating exchange rates.
- Main implementation challenges (as stated):
  - data from wholesale market often not timely nor fully representative; liquidity issues in emerging markets
  - market fragmentation when order flows concentrate among a few institutions
  - choice of methodology (average, weighted, median, quotes, transactions) affects outcomes and must balance accuracy, transparency, and relevance
  - weak institutional frameworks increase manipulation risk
  - technological integration and infrastructure needs
  - need for ongoing maintenance and periodic review
- Policy interaction: central bank FX policy responses highlight the need for transparent calculation methodologies.

### Framework Design — Main Design Models
- Four key models described:
  1. Central bank-managed fixing sessions (used in cases of extreme market dysfunction; similar to interbank FX auctions; may involve pre-submission of bids and offers; can produce a market-clearing midpoint rate; now rarely used).
  2. Simple arithmetic average of FX spot quotes submitted by a panel of most active domestic banks (quotes are indicative; often trimmed of extreme values; treats contributions equally and may introduce bias; limited recent use).
  3. Simple arithmetic average of FX spot quotes (indicative or executable) observed at a point in time or over a short time window on most active trading platforms (suitable where continuous trading exists; critical to observe tradable rates not indicative; can be published during the trading session; widely used in emerging markets and by benchmark providers in advanced economies).
  4. Weighted arithmetic average of FX spot transactions observed over a short time window or the entire trading session on the most active trading platform (most sophisticated; suitable where transactions are observable).
- Advantages and drawbacks highlighted:
  - Fixing sessions: enable endogenous demand/supply discovery in absent interbank markets; risk that opening rate may not be equilibrium; sometimes no trading occurs—then interbank quotations may be used.
  - Panel quotes: supportive in early sessions for price guidance but representativeness not ensured; trimming used to reduce outliers.
  - Quasi–point-in-time platform observations: lower manipulation risk if market deep and heterogeneous; manipulation risk remains if participants can concertedly influence rates at the observation point.
  - Session-long transaction observation: robust to intraday liquidity fluctuations and manipulation attempts; earliest publication is after session end or next business day, which can reduce representativeness and may conflict with IMF Guidance on Multiple Currency Practices.

Box 1 — IMF Guidance for Official Exchange Rates under the Multiple Currency Practices Policy
- Context: in less-developed FX markets authorities often use one-day-lagged official exchange rates for FX transactions when current-day data are unavailable.
- Conditions under which one-day-lagged rates will not result in MCP:
  - the lagged official exchange rate is determined as a weighted average of transaction exchange rates;
  - when a wholesale FX market exists, calculation is based on its transaction rates;
  - if no wholesale FX market or its rates are not representative, compute using weighted average of transaction exchange rates between intermediaries and their clients, or a combination of both;
  - transactions included must have occurred on the business day immediately preceding the day on which the official exchange rate is to be used.
- Example provided (as stated):
  - central bank announces official exchange rate (R/USD) at 16:00 every trading day;
  - official exchange rate is determined by the weighted average interbank market exchange rates from 15:00 on the previous trading day (T) to 15:00 on the current trading day (T+1);
  - government trades with the central bank the next day (T+2) using this official rate calculated on the current day (T+1);
  - because the official exchange rate used on T+2 includes transaction information from two days before (the transactions between 15:00 and 17:00 on T), the calculation does not meet the criteria for a one-day-lagged exchange rate that would avoid giving rise to an MCP.
- Monitoring requirement: the example indicates the need to monitor such rates when used for government transactions.

### Framework Design — Key Principles
- Three foundational principles:
  - Robustness: calculate and disseminate the reference rate every business day without significant operational risks, even during adverse market conditions; highest priority.
  - Representativeness: reference rate should reflect prevailing market conditions formed by competitive demand and supply; use only transactions where counterparties’ bargaining power is sufficiently similar.
  - Replicability: market participants should be able, on average, to execute FX transactions at rates sufficiently close to the reference rate without prior knowledge of its value; transparency and market-based inputs facilitate replicability.
- Trade-offs:
  - Including only larger transactions improves representativeness but harms robustness due to infrequency in thin markets.
  - Expanding eligibility to dealer-to-customer transactions and smaller trade sizes improves robustness but can undermine representativeness.

*Foreign Exchange Reference Rates — IMF | MCM Technical Assistance Handbook (chapter content provided).*

### Box 2. IOSCO Principles

### Box 2. IOSCO Principles

### Overview
- The IOSCO Principles for Financial Benchmarks, published in July 2013, outline 19 principles for benchmark administrators to follow, grouped into four categories that aim to enhance the reliability, transparency, and integrity of financial benchmarks.

### Governance (category)
- Comprises five principles ensuring appropriate arrangements to protect the integrity of the benchmark determination process.
- Key provisions:
  - Addressing conflicts of interest.
  - Requiring Administrators to retain primary responsibility for all aspects of the benchmark determination process.
  - Establishing oversight functions to review and challenge the benchmark determination process.
  - Ensuring benchmarks are administered in a transparent and accountable manner.

### Quality of the Benchmark (category)
- Comprises five principles promoting quality and integrity of benchmark determinations.
- Key provisions:
  - Benchmarks should be based on prices, rates, indices, or values formed by competitive market forces.
  - Addresses data sufficiency and the hierarchy of data inputs.
  - Ensures transparency of benchmark determinations and clarity on how benchmarks are calculated and disseminated.

### Quality of the Methodology (category)
- Comprises five principles focusing on the quality and integrity of methodologies used.
- Key provisions:
  - Requires publication of minimum information about the methodology.
  - Addresses procedures for material changes to the methodology.
  - Establishes policies for benchmark transitions or cessations.

### Accountability (category)
- Comprises four principles with mechanisms to ensure benchmark Administrators are accountable.
- Key provisions:
  - Establishing complaints procedures.
  - Mandating audits and audit trails to maintain records of benchmark-related activities.
  - Ensuring cooperation with regulatory authorities.

### Objectives and Constraints at Various Levels of Market Development
- The IOSCO Principles address application to both active and illiquid markets.

- Active markets:
  - Defined by observable bona fide, arm’s-length transactions with considerable size, liquidity, many participants, or limited concentration.
  - Benchmarks should ideally be based on prices formed by competitive forces of supply and demand, anchored by observable transactions.
  - Individual determinations need not rely solely or predominantly on actual transactions.

- Illiquid or fragmented markets:
  - Recognize limitations due to lack of observable transactions.
  - Benchmarks may rely on alternative data sources or methodologies (e.g., expert judgment or indicative quotes) if transparent and robust.
  - Principles are applied proportionally to avoid undue regulatory burdens.

- Factors considered in determining market maturity:
  - Interbank and Interdealer FX Market Activity.
  - Official Intervention (degree inversely related to market development).
  - Dominance of Private FX Inflows.
  - Sophistication of Derivatives Markets.
  - Market Infrastructure and Capacity.

- Market-stage implications:
  - Nascent Markets:
    - Factors largely missing; no active market for foreign exchange.
    - Inputs predominantly trades between banks and FX bureaus with their clients.
    - Benchmark calculated only once per day, based on deal flow across a full business day, made available on an ex-post basis, primarily for accounting purposes.
    - Where official flows are critical, benchmark may be derived from official auctions or fixing sessions.
    - Benchmark described as robust, moderately representative, but lacking replicability.
  - Emerging, Developing and Maturing Markets:
    - Improvements in maturity factors; emergence of an active market.
    - Transactions typically encompass interbank trades and wholesale deals executed by large bureaus with clients and banks.
    - Benchmark role expands to portfolio valuation and performance measurement; replicability becomes increasingly critical.
    - Exchange rates published at regular intervals during the day, capturing activity within specific time windows.
    - Robustness and representativeness maintained by including all transactions within each window and applying volume weighting; firm quotes from dealers may be incorporated.
    - Further improvements in market depth may be needed to establish replicable benchmarks.
  - Established or Mature Markets:
    - Meet all maturity factors and have active markets.
    - Market activity primarily driven by interbank and interdealer trades and electronic trading channels.
    - Benchmark computed over short, specific time windows—merging point-in-time data with a narrow window—and typically include mechanisms to filter out outlier or unrepresentative transactions.
    - Supports development of derivative hedging instruments, index construction, portfolio valuation, and performance measurement.
    - Benchmarks balance replicability, representativeness, and robustness.

### Timing, Sample Size, Number of Rates Published
- Sample window sizing evolves with market sophistication; as markets deepen, windows can be narrowed to representative trades and liquid currency pairs.
- Regulatory business-hour definitions may limit overall window size.
- Central banks can identify shorter windows or point-in-time observations (e.g., market opening, midday, closing) for regular publication.
- Length of fixing window drives client transaction costs and dealer profitability:
  - Longer windows: reduce client execution costs, spread hedging activity, lower price impact; reduce risk-adjusted profitability for dealers.
  - Shorter windows: increase dealer returns but raise client costs.
  - Differences in weighting function within the window play a comparatively minor role.
- Administrators should weigh trade-offs between client costs and dealer participation; tailor fixing designs to market structure (Oomen and Muhle-Karbe, 2024).
- Multiple exchange rates:
  - Should be market-driven, consistent, and transparent.
  - Number of rates per objective should be limited (e.g., only one closing rate per day).
  - Methodologies should be consistent, comparable, and fully documented.
  - Typically, central bank offers a single official exchange rate for a specified period, complemented by additional indicative rates.
- Publication timing:
  - Should balance dealing community needs and wider public needs.
  - Banks, FX cash offices, asset managers, corporate treasurers may require higher-frequency indicative rates (opening, midday, closing).
  - Public may prefer on-demand indicative rates or average and closing rates.
- One-day lag publications:
  - When data submission delays occur, rate may be computed and published with a one-business-day lag.
  - Derived exchange rate in such cases should be used solely for accounting purposes and communications must state this limitation.
- Participation and data submission:
  - Central bank should specify data requirements, identify submitting institutions, ensure infrastructure, and govern submissions under rules consistent with IOSCO Principle 14 on submitter code of conduct.
  - May mandate data collection through regulatory requirements imposed on licensed market agents.
  - Distinguish between submission of indicative quotes and reporting of transactions; transaction reporting requires greater infrastructure.
- Publishing a single mid-rate as official reference rate is a best practice:
  - Mid-rate reflects fair market value without embedded transaction costs.
  - Buying and selling rates may be disseminated but their roles must be clearly defined.
  - Official rate should be published solely as a mid-rate to maintain objectivity and market utility.

### Eligible Transactions and Quotes
- Eligibility criteria critical for representativeness and robustness:
  - Transactions must involve same financial instrument (FX spot transactions) and ideally be sourced from a single marketplace where prices are formed by competitive forces (wholesale interbank transactions).
  - Criteria may be relaxed under predefined contingency conditions when wholesale market inactive.
- When wholesale market not active:
  - Scope may expand to dealer-to-customer trades.
  - Determine transaction amount threshold; excluding very small retail transactions may be appropriate.
  - Minimum transaction amount threshold can be based on market practices, minimum lots on electronic platforms, and historical data analysis.
- Considerations for dealer-to-customer trades:
  - Dealer-to-customer transactions occur with a spread to wholesale market; mid-rate may tilt when supply/demand imbalances exist.
  - Reporting of dealer-to-customer transactions may be manual and prone to errors, corrections, and delays.
- Data quality:
  - Typically better for interbank transactions; retail transactions often operationally difficult and low volume share—justifying exclusion.
  - In countries with strict capital controls, retail prices can deviate substantially from wholesale market.
- Spot convention:
  - Spot (T+2) most commonly used; some markets use T+0 or T+1.
  - Typically one convention should define eligible transactions; using all three requires recalculation into spot using interest rate differential.
  - Higher interest rate differential increases measurement error.
- Exclusions:
  - Transactions between central bank and market participants, and between central bank and government, should be excluded.
  - Such transactions often conducted at previous day's official rate and not determined by competitive forces.
  - Exclusion justified where central bank activity accounts for substantial share of interbank activity.
- Very large transactions:
  - Can have outsized impact on volume-weighted average reference rate; may relate to government operations, SOEs, or debt conversions.
  - Solutions: use median volume-weighted rate or apply thresholds to exclude very large transactions.
- Spot legs of swap transactions should not be considered eligible transactions.

### Central Tendency Estimators
- Choosing the measure of central tendency is critical for transaction-based FX benchmarks.
- Unsuitable estimators (e.g., simple mean with skewed data) can produce biased or unrepresentative benchmarks.
- Robust alternatives include weighted average, median, or trimmed mean depending on market depth, transaction-size distribution, and manipulation risk.
- Comparison between weighted median and weighted average across five features:
  - Consistent: Weighted Median — Consistent; Weighted Average — Consistent.
  - Robust to outliers: Weighted Median — Highly robust; Weighted Average — Not robust.
  - Unbiased: Weighted Median — Unbiased (under certain conditions); Weighted Average — Typically unbiased.
  - Distribution assumptions: Weighted Median — Less dependent on assumed distribution; Weighted Average — More dependent on assumed distribution.
  - Efficient: Weighted Median — More efficient in terms of absolute deviation; Weighted Average — More efficient in terms of variance.
- Conclusion: Weighted median is superior across most features; weighted average is optimal when parametric distributional assumptions hold.

### Data Sufficiency Criteria and Contingency Calculation Methods
- Data sufficiency derived from principle requiring prices be driven by competitive forces in arms-length transactions.
- Data insufficient if criteria not met; criteria must be calibrated per country reflecting central bank preferences between robustness and representativeness.
- Even well-developed markets can have episodes of insufficient data during financial stress.
- First step: define central bank’s tolerance for insufficient data expressed as percentage of trading days expected to be insufficient; estimate using at least one year of historical data and re-estimate annually.
- Data insufficiency triggers:
  - Number of transacting banks on a trading day falls below a set threshold.
  - Daily volume is less than a specified percentage of the average daily volume.
  - High concentration of transactions where a few banks account for a substantial share of turnover.
- Rules of thumb:
  - Percentage of trading days with highly concentrated market turnover—defined as days when the top three banks account for more than 80 percent of turnover—should be lower to indicate greater robustness.
  - Percentage of trading days with zero or near-zero trading volume should ideally be as close to zero as possible.
  - Percentage of trading days with very few active market participants (i.e., fewer than five banks) serves as an important metric.
- When interbank market is inactive, central bank should rely on alternative data sources in a predefined hierarchy:
  - Dealer-to-customer trades (if not already used).
  - Request for Quote (RFQ), or executable or indicative quotations (fixings based on snapshots at specific times or multiple quotations during trading hours).
  - Transactions with different currency pairs with pre-determined guidance to recalculate cross rates.
  - Previous day's benchmark as an input.
  - Expert judgment (allowed by IOSCO Principle 8 but may undermine credibility and automation).
- Alternative data sources supplement eligible transaction data until sufficient data is available, following top-down input hierarchy.

### Transition
- Central banks may adjust calculation methodology or benchmark framework when market development improves.
- Ideally executed in a single step; conceptually preferable and may include retroactive calculation of the new benchmark (retroactive calculation may be prevented by lack of historical data).
- Many central banks prefer gradual changes to avoid excessive volatility, revaluation effects, and impacts on expectations—especially where managed exchange rate regimes exist.
- Gradual transition approach:
  - Compute weighted average of legacy rate and new rate with drifting weights (legacy rate weight starts at 1, new rate at 0, reversed by end).

### Governance (administrator role and central bank advantages)
- Administrator and other governance roles can be performed by various entities, but central bank is best positioned for FX benchmarks.
- Benefits of central bank administration:
  - Avoiding Conflicts of Interest: Central banks can administer benchmarks independently, eliminating conflicts when private entities manage benchmarks and offer linked financial products.
  - Enhanced FX Market Functioning: Central banks have interest in improving FX market functioning and supporting better monetary policy transmission.
  - Credibility and Market Insight: Central banks are credible government institutions with deep understanding of market conditions; central-bank-administered FX benchmarks are trusted and used by entities including government institutions for debt operations, export-import transactions, and balance sheet revaluations.

*Foreign Exchange Reference Rates — IMF | MCM Technical Assistance Handbook*

### Box 3. IOSCO Benchmark Governance Framework

### Box 3. IOSCO Benchmark Governance Framework

### Key roles in the benchmark governance framework
- Administrator
  - Responsible for managing the overall benchmark process, ensuring the integrity of the benchmark, and facilitating communication with stakeholders.
- Administrator Oversight Function
  - An independent committee that establishes and reviews the benchmark's methodology, monitors the benchmark governance process, and addresses potential conflicts of interest.
- Calculating Agent
  - Tasks: accurate calculation of benchmark values based on provided data; processing and analyzing collected data in accordance with the established methodology.
- Publishing Agent
  - Tasks: disseminate the calculated benchmark values to the public and relevant stakeholders; ensure published information is accurate, timely, and compliant with defined standards.
- Audit Function
  - Tasks: review and assess processes and methodologies used in benchmark calculation and governance; ensure benchmarks are calculated accurately and transparently and procedures are followed.
- Note: The roles of the calculating agent, publishing agent, and audit may be fulfilled by entities other than the administrator and its oversight function.

### Central bank as administrator — assignment and internal design
- Responsibility for all aspects of the benchmark determination process can be assigned to a functional unit that:
  - Operates in the FX market or possesses up-to-date information on market developments and intelligence.
  - Initiates benchmark establishment, communicates with market participants and central bank management, coordinates internal processes, and prepares internal and external documents and regulations.
- The central bank must establish an oversight function (e.g., an Oversight Committee, OC) to identify, mitigate, and manage conflicts of interest and develop procedures and documents regulating OC activities:
  - OC terms of reference; criteria for selecting/electing OC members, head and deputy, replacements; declarations of any conflicts of interest by OC members.
- OC composition and engagement:
  - OC should include representatives from the FX market community, selected by professional FX traders' associations, such as ACI, or directly by the treasuries of licensed banks.
  - It is recommended that the central bank maintain a majority in the OC's composition, with a representative from the central bank as the head.
  - The OC may develop the benchmark methodology and recommend it for approval by the central bank's decision-making body (e.g., Governing Board).
  - It is advisable for the central bank to engage with market participants prior to benchmark introduction.

### Oversight Committee (OC) tasks and responsibilities
- Periodic review and assessment of whether the existing benchmark definition or methodology accurately reflects the underlying market (examples: market volume shifts to a different settlement convention such as tom/next to spot, or to a different currency pair).
- Review proposed methodology amendments and ensure necessary remedial actions are implemented when market structure changes.
- Review and recommend procedures for termination of the benchmark to the central bank's decision-making body.
- Verify adherence to the existing methodology, consider audit results, and monitor implementation of remedial actions to ensure integrity and effectiveness of the control framework.
- Review cases where expert judgment or contingency conditions were applied (non-standard procedures for benchmark production).
- Review and address complaints received by the central bank regarding benchmarks.

### Separation of duties and operational practice
- Daily benchmark determination process must be separated among business areas:
  - Central bank: data collection, validation, and conducting the data insufficiency test.
  - Other structural units: fulfill roles of Calculation and Publishing Agents.
- Good practice: Administrator and Calculation Agent should come from different business areas and report to different deputy governors to mitigate concentration of roles and conflicts of interest.
- Benchmark calculations:
  - Preferably performed automatically using an internal application and without discretion.
  - When using contingency methodology, a predefined methodology must be applied and management approval obtained before dissemination.
  - Contingency procedure should ensure timely receipt of additional data (e.g., firm quotes) to complement actual transactions used for standard calculation.

### Accountability and transparency (J)
- Transparency and adherence to IOSCO principles improve accountability, credibility, and wider adoption.
- Recommended benchmark disclosures on the central bank’s website:
  - A methodology statement including benchmark definition, publication time, standard and contingency computation methodologies, data insufficiency triggers, and error handling.
  - Disclosure of cases when the contingency methodology is used.
  - Contact email for complaints regarding the benchmark methodology; complaints can be considered during periodic reviews. Complaints regarding market conduct should be escalated to the relevant authority.
- Audit arrangements:
  - Appoint an internal or external auditor to periodically review and report on adherence to methodology.
  - For many emerging markets’ central banks, an internal auditor may be sufficient given size, market development, and cost considerations.
  - Consider publishing a statement of compliance with “IOSCO principles for Financial Benchmarks” and appointing an external auditor to provide periodic assurances on this compliance statement.
- Complementary publication:
  - Publish aggregated indicators on market turnover to help users assess representativeness and inform clients about market liquidity (helpful for scaling FX orders in line with market turnover).

### Prevention of market manipulation (K)
- Risks and drivers:
  - Fixing orders (currency conversions executed at a certain benchmark exchange rate) create open positions and can provide non-public information on potentially market-moving flows, creating incentives and capabilities for manipulation.
  - Point-in-time observation methodologies are more vulnerable if not robust to concentrated market activity.
- Detection methods:
  - Supervisory methods: privilege of law enforcement agencies; collect evidence on collusion (e.g., monitor bilateral communications).
  - Observatory methods: analyze outcomes of possible collusion — trades (volatility and trading volume), quotes (volatility), contributions (distribution).
  - Observational difficulty: distinguishing between elevated market activity and manipulation; select sensitive dates (month-end, quarter-end, year-end) and times (data release/announcement schedule) for monitoring.
  - Monitor market participants’ response to benchmark design changes to assess effectiveness of policy actions.
- Prevention strategies:
  - Align benchmark design with usage to reduce manipulation risk.
  - Options:
    - Make the benchmark robust to manipulation efforts (e.g., lengthen observation window).
    - Eliminate motivation to manipulate (e.g., force banks to cease accepting fixing orders via moral suasion or regulation).

### Cross rates (L)
- Definition: exchange rates between two currencies not involving the dominant currency pair (typically against the U.S. dollar under direct quotation).
- In this handbook context: cross rate is the value of the local currency against another currency that is not the dominant currency pair and is therefore not readily available from frequently traded domestic pairs.
- Derivation typically uses an exchange rate of a common currency (typically the USD).
- Uses: facilitate international transactions without USD, support investment decisions, manage currency risk, identify arbitrage opportunities.
- Practical considerations:
  - Globally convertible currency pairs often derived from international fixings (example: WM/Reuters 4 pm London Fix).
  - For illiquid cross pairs, central bank rates are often indicative; the central bank does not trade at those rates.
  - Use of mid-rate for valuation and accounting recommended; communicate methodology, timings, and relevance.

### Technical assistance (IV) — recommended TA approach
- Core steps:
  - Map out the relevant objectives and constraints.
  - Determine the design parameters of the methodology.
  - Design a data collection process, if needed.
  - Draft a public policy document.
- A. Map Out the Relevant Objectives and Constraints
  - Perform quantitative and qualitative evaluation of FX market development; quantitative assessment should use granular, ideally transaction-level, data.
  - Assess importance of market segments (foreign exchange bureaus, dealer-to-customer, interdealer) and the parallel market (volume, spread, participants) via surveys/anecdotal evidence.
  - Survey established market participants (market makers and clients) on qualitative and quantitative indicators of market functioning.
  - Explore potential and actual use of reference rates via review of legal/regulatory framework and interviews with central bank subject matter experts and stakeholders.
  - Set objectives and constraints to interpret and prioritize principles of representativeness, reliability, and replicability.
- B. Determine the Design Parameters of the Methodology
  - Select the reference rate model most robust and representative given market conditions and willingness to rely on contingency methodologies.
    - Fixing session model: appropriate where interbank FX markets are structurally absent and active price discovery is needed.
    - Panel-based quote model: suitable for markets with some interbank activity but limited transaction data.
    - Point-in-time observation model: best for continuous, liquid trading with reliable executable prices.
    - Full-day transaction-weighted model: appropriate for intermittent or shallow liquidity to capture full range of transactions.
    - Hybrid approaches may be warranted at intermediate market development stages.
  - Back-test historical data to set parameters: scope of eligible transactions, transaction size thresholds, sample window, central tendency estimators, choice to publish mid-rate only or also buy/sell rates, data sufficiency criteria, and contingency prioritization.
  - Objective: balance representativeness, robustness, and replicability rather than maximize stability.
- C. Design Data Collection Process If Needed
  - Use a common reporting template across reporting entities to standardize granular data enabling filtering and cross-checking (Appendix I example).
  - Ensure data management includes access control, audit trail, and data retention to distinguish legitimate data cleaning/error handling from manipulation.
- D. Produce a Public Policy Document
  - Public policy document should outline governance and methodological arrangements, explain how the main decision-making body exercises responsibility, be comprehensive enough to enable effective audit, and include error handling, communication procedures, and stakeholder consultation arrangements.

*Italic: Box 3. IOSCO Benchmark Governance Framework — content extracted from the provided source.*

---


_Source: https://www.imf.org/-/media/files/publications/miscellaneous/english/2025/ta-handbook-chapter-foreign-exchange-reference-rates.pdf_