MANAGEMENT OF FOREIGN CURRENCY RESERVES

Risk Management

Risk Management Framework

It is generally well accepted that a central bank should adopt the most rigorous risk management practices in respect of all aspects of its reserves management processes. At the most general level, this is reflected in the need for the complete separation of functions between the Front, Middle and Back Offices, the Audit Department, the Accounting Department and the Risk Management Unit (Figure 1). Separation of these functions is designed to ensure that exposures are within established limits and to minimise the opportunities for fraudulent behaviour. While there is not a unique ‘blueprint' for this framework, a number of basic elements should be considered:

• At the very highest level, the decision-making hierarchy for the investment of reserves should be clearly defined. This hierarchy would normally be established by the Governor or Board of Directors and would include the overall objectives of reserves management, the specification of the investment mandate and would identify who is responsible for implementing this mandate.
• To reduce the risk of fraud, the separation of functions must be transparent and enforceable. Most importantly, there should be complete separation of those who initiate transactions (Front Office) and those who arrange the settlement of transactions (Back Office). Depending on a central bank's particular circumstances, this separation may not be readily achievable but at a very minimum must be controlled by restricting staff access to sensitive systems (for example, SWIFT).
• The separation of functions extends beyond the separation of the transaction and settlement functions. Just as important is the separation of the responsibility for managing financial exposures (Front Office) and measuring them (Middle Office). Specifically, a Middle Office would be responsible for setting (and reviewing) operational guidelines and limits, would be responsible for performance measurement and attribution and should be responsible for ensuring the continued relevance of the benchmark.
• Notwithstanding the role that the Middle Office plays as an on-line auditor, provision should also be made for regular independent audits of the reserves management process. 

An important element of any risk management framework is that the management policies and procedures are clearly understood by all staff involved in the process. This is best achieved by documenting all aspects of the investment mandate as well as all operational controls in a single document. Specifically this documentation should include:

• The reporting lines and responsibilities of all functional areas involved in the management of foreign reserves.
• All limits and controls extended to portfolio managers.
• Details of the performance benchmark and systems used for performance attribution.
• The list of all authorised instruments and any limits that may apply to them.
• The eligibility criteria for the selection of trading counterparties.
• The framework for determining the maximum credit exposures permitted with each counterparty.
• Details of the methodology for measuring risk exposures (both market risk and credit risk).
• Clear procedures for notifying senior management of limit breaches or any other ‘exceptional' events relating to the reserves management process.

Procedures also need to be established for how changes can be made to the investment mandate such as a proposal for the introduction of a new instrument or a change in the composition of the benchmark. These procedures need to include which level of senior management is authorised to approve changes and which functional areas need to be consulted before changes can be implemented.

Risk Management Tools

At a more operational level, there are a variety of possible risk management tools available to central banks. Explicit in the design of the foreign currency benchmark is a decision about the acceptable level of market, credit and currency risk that a central bank is prepared to accept in light of its investment objectives and investment universe. In addition to this, central banks may give their portfolio managers latitude to deviate from these benchmarks. This discretion can be expressed in many ways – as variations in duration, variations in portfolio composition or variations in currency shares. A challenge for central banks is to quantify these risks on a consistent basis and incorporate these measures in the limit processes.

At a very basic level, the likely return (dollars-at-risk) from portfolio manager exposures can be readily approximated from the duration of the actual and benchmark portfolios. As a general rule, if the duration of the actual portfolio is one year longer than the duration of the benchmark portfolio, this would represent a dollars-at-risk exposure equal to one per cent of the value of the portfolio for every 100 basis point change in yields. In this case it is relatively easy to nominate a trading limit based on a maximum acceptable dollars-at-risk exposure.

The attractiveness of a dollars-at-risk measure is its simplicity. Unfortunately, this also represents its biggest limitation as it fails to capture the effect of changes in the shape of the yield curve (it assumes that the yield curve move is parallel) and it fails to capture the unique price sensitivity of individual instruments (convexity). To some extent, this deficiency can be overcome by granulating the limit by, for example, specifying dollars-at-risk limits at a more disaggregated level such as by maturity band or even by instrument.

Another related deficiency of dollars-at-risk measures is that they are not necessarily additive across portfolios as they make no allowance for price volatilities and the correlation in price movements between securities and between portfolios. For example, although US and European yields often move in the same direction, the size of the moves will not necessarily be in the same magnitude. This means that in the event of a global rally in yields, the gains on a long US security position may not completely offset the loss on a short European security position. This widely recognised deficiency has, to some extent, been overcome by the increased reliance on Value at Risk (VaR) measures.4 Although VaR measures have a number of advantages (most notably that they enable exposures across a variety of instruments and portfolios to be aggregated into a single number), the implicit assumption about the normal distribution of asset returns and the use of historical price relationships mean that they are likely to be poor estimates of portfolio returns in the event of extreme market movements.

The VaR and dollars-at-risk measures tend to focus on more narrowly measuring the market risk of portfolio exposures. Neither measure is particularly good at quantifying the risk that an issuer of securities or a counterparty to a transaction will fail to meet their financial obligations at maturity or at settlement. While variations in the probability of default is implicit in the price of all securities and is therefore captured in the VaR and dollars-at-risk measures, there are many dimensions to measuring and managing credit risk that a central bank needs to consider:

• Credit exposures to particular issuers can be controlled by limiting the range of eligible securities to those of highly-rated issuers. Where investment mandates permit investment in less highly-rated instruments (such as corporate securities) individual instrument or issuer limits may be required.
• Individual counterparty limits tend to have two dimensions – the level of capital that a central bank is prepared to expose to a particular counterparty and the maximum potential financial loss associated with each instrument type. In the case of the acceptable capital exposure, this tends to be a function of the credit rating of the counterparty and its capital base (or the capital base of the central bank) while the maximum instrument-specific exposure is based on the likely financial cost of any counterparty failing to meet its financial obligations in regards to that instrument.5

Derivative Risk Management

Trading derivative instruments is generally perceived to be more risky than trading physical securities. This perception tends to be more a reflection of bad risk management practices in the past than a reflection of the products themselves. At the very basic level, a futures contract has essentially the same market risk characteristics as the underlying instrument and, in the case of exchange-traded derivatives, virtually all of the counterparty credit risk exposure is eliminated by the clearing house. In these simple cases, central banks should be relatively relaxed about using derivative products so long as the exposures are fully reflected in the balance sheet and they are not used to gear the portfolio beyond the value of the portfolio. On the other hand, derivatives which have non-linear payment characteristics (such as options) require specialist knowledge and complex risk management systems that are likely to exceed the resources available to many central banks.

Operational Risk Management

Aside from the direct financial risks associated with managing a portfolio of assets, central banks also face a significant range of procedural or operational issues that pose a significant reputational exposure to the institution. Examples of these operational risks include:

• failure to deliver on contractual obligations resulting from trading and settlement system failures;
• ineffective backup arrangements in the event of a contingency event;
• failure to prevent fraudulent behaviour or to prevent excessive risk taking (including executing transactions in ineligible securities or with unauthorised counterparties).

Operational risks are fairly easy to identify and can often be eliminated through system enhancements and well documented procedures. Identifying and, more importantly, eliminating these risks isn't costless, however. Introducing new trading and settlement systems, for example, may involve a significant commitment of staff that were previously involved in the reserves management process and may take several years to fully implement.

Footnote

4. A VaR measure estimates the maximum expected dollar loss on a given position (or portfolio) over a given holding period with a particular level of confidence. Alternatively, it can be described as the potential loss that is expected to be exceeded on a given number of occasions (determined by the confidence interval). The holding period reflects the time it takes to liquidate, neutralise or reassess a position and, as such, should reflect the nature of the portfolio. A daily holding period is appropriate for institutions which trade highly liquid instruments. A one-week holding period may be more suitable for institutions which adjust their portfolios less frequently or trade illiquid instruments. (back to text)
5. A risk factor on a deposit would be significantly higher than the risk factor applying to a fully collateralised loan. In the case of the latter, the expected loss is restricted to any forgone interest and any possible collateral shortfall (which can be minimised with expedient use of haircuts and margin calls). Similarly, the financial loss on a foreign exchange swap prior to settlement is limited to the movement in the respective exchange rate. That said, on the settlement date, the replacement cost increases sharply in the circumstance that one leg of the transaction settles well before the other. (back to text)