• Action by individual banks to control their FX settlement exposures
• Action by industry groups to provide risk-reducing multi-currency services
• Action by central banks to induce rapid private sector progress

This section describes the three elements of the strategy in detail and discusses their key implications from a central bank policy perspective.

4.1 Action by individual banks to control their foreign exchange settlement exposures

Individual banks should take immediate steps to apply an appropriate credit control process to their FX settlement exposures. This recognises the considerable scope for individual banks to address the problem by improving their current practices for measuring and managing their FX settlement exposures.

4.1.1 Description of recommended action

Improve practices. Individual banks could improve their settlement practices so as to gain better control over their FX settlement exposures. In particular, banks could improve their back office payments processing, correspondent banking arrangements, obligation netting capabilities and risk management controls sufficiently to permit them to:

• Measure FX settlement exposures properly
• Apply an appropriate credit control process to FX settlement exposures
• Reduce excessive FX settlement exposures for a given level of trading

Measure exposures. First, banks could adopt internal procedures that would permit them to measure their FX settlement exposures properly. For instance, a bank could develop a system that frequently updates its current and future global exposures as it executes new trades and as unsettled trades move through the settlement process. This would give it much more accurate and timely information regarding its FX settlement exposure. This capability, however, might not be immediately feasible, particularly for an international bank actively trading a wide range of currencies with a substantial number of counterparties out of many locations without the benefit of a consolidated risk management system. Nevertheless, such a bank (or, at least, each of its trading centres) could adopt procedures to update its exposure calculations periodically (e.g. once or twice a day) and to measure its minimum and maximum exposure at any moment on the basis of all available information. In either case, Appendix 1 provides guidelines that a bank (or each of its trading centres) could use to measure its current and future exposures.

1. The effect on monetary policy implementation: monetary policy implementation could be affected by the impact ... on the ability of the central bank to control the supply of and to forecast the demand for reserve balances, and by the impact on open market operations, central bank lending and other operating procedures. This might affect interest rates and exchange rates. 2. The adequacy of private sector sources of liquidity to support settlement in each currency: this could be influenced by the availability during settlement of deep and liquid money markets, of final transfers into settlement accounts and of collateral to support funding transactions. 3. The impact on systemic risk: this could depend on the effect ... on private sector motivation to design new methods to reduce settlement risks, on the ability and incentive of the public and private sectors to manage credit and liquidity risks, and on the degree of reliance on public and private sector credit and liquidity. 4. The well-founded legal basis of settlement arrangements and entities: this would depend in part on the legal status of settlements in each country and on the legal implications of the location and corporate form of settlement entities. 5. The likely competitive effects in private financial markets: this would depend on the markets to be served, on the participants and entities that would benefit from access to ... services and on likely changes to correspondent banking relationships. 6. The cost-effectiveness ... from the private sector perspective: this would reflect initial investment costs and the implementation timetable, the ongoing operating costs relative to the status quo and the costs of any required idle balances that might arise as a result of prefunding of debits or delayed access to credits. 7. The acceptability ... from an individual central bank perspective: this would reflect initial investment costs (e.g. the cost of new technology) and the implementation timetable; ongoing operating costs; required legislative and policy changes; implications for central bank supervision or oversight; implications for the role of the central bank as liquidity provider; likely shifts in the loci of financial activity; and the required degree of coordination, cooperation and sharing of confidential information. Source: Central Bank Payment and Settlement Services with respect to Cross-Border and Multi-Currency Transactions, Basle, September 1993.

Manage exposures. Second, a bank could adopt internal procedures for explicitly assessing the risks and rewards of its FX settlement activities, thereby permitting it to manage its properly measured exposures on the basis of fully informed business judgements. As part of an effective management approach, a bank could choose to control its properly measured FX settlement exposures in a manner consistent with the way in which it controls its other credit exposures. For instance, many banks currently set a limit on their total credit exposure with a single counterparty based on an internal credit analysis. Such a limit would generally apply to all operations that generate credit exposure, whether a loan, a deposit, a letter of credit or any other formal extension of credit. Some banks also set separate sub-limits on different possible durations of credit exposure (e.g. remaining exposures of up to 7 days; up to 30 days; up to 90 days; etc.). Furthermore, some banks that have many offices around the world but do not have a global real-time limit monitoring system divide each limit or sub-limit among the various entities and monitor them on a decentralised basis. This control process enables a bank (or a particular office) to undertake any combination of credit-generating activities with a single counterparty and still assure senior management that the bank's overall credit exposure will remain within the level it considers appropriate.

This assurance, or any similar assurance that could be provided by other effective credit control processes, could be extended to credit exposures that arise in settling FX trades simply by including properly measured FX settlement exposures under the same set of controls. For this to work effectively, however, a bank would need to accept the proposition that - when dealing with a particular counterparty - FX settlement exposure represents the same credit risk, and the same probability of loss, for the bank as, for example, a loan of identical size and duration. Once a bank applies its standard credit controls to FX settlements, it could assure itself that these exposures would not exceed a level the bank considers appropriate.

Reduce excessive exposures. Third, even without lowering the scale of its FX trading, a bank could reduce any FX settlement exposure it deems excessive and decrease the uncertainty surrounding the size of its exposures by improving its settlement practices. For instance, by eliminating overly restrictive payment cancellation deadlines and shortening the time it takes to identify the final and failed receipt of bought currencies, a bank could lower its actual and potential FX settlement exposure for the same level of FX trading. Depending on a bank's trading pattern, the use of available bilateral or multilateral obligation netting arrangements could reduce exposures even further. If necessary, in certain cases a bank may further protect itself against excessive FX settlement exposures by, for instance, requiring collateral from its counterparties.

4.1.2 Central bank policy perspective

Reduced FX settlement risk. Overall, by improving its settlement practices, a bank could gain more effective control over its FX settlement exposures in settling trades with any counterparty in any currency. This would permit a bank to protect its financial health and to reduce its reliance on potentially destabilising actions at times of market stress. Furthermore, a bank could take immediate steps to improve its practices.

Informed credit judgements. This approach to reducing risk makes use of the traditional strength of individual banks in reaching informed credit judgements and in pricing and controlling credit risk properly. Today, many banks are not aware that they can incur sizable FX settlement exposures overnight and during weekends and holidays. Once a bank fully understands and quantifies its FX settlement exposures, it could apply a rigorous risk/reward analysis to them and conclude that a reduction in exposures for given trading levels is in its economic interest. For this purpose, a bank could use an approach that is generally consistent with its approach to the control of its counterparty credit exposures in other markets, and would therefore create no new conceptual difficulties.

Reduced sources of systemic risk. An improvement in settlement practices would also help prevent a bank from either over-reacting or under-reacting to changes in the credit quality of its counterparties. For instance, overly restrictive payment cancellation deadlines and excessive delays in identifying final and failed receipts leave many banks today with three poor options when faced with a sudden increase in concern about a counterparty: continue to process their outgoing payments as usual (a possible under-reaction); attempt to monitor the flow of particular incoming receipts and outgoing payments on an ad hoc, exceptional basis (a potentially unreliable and expensive process); or halt all trading and payments (a possible over-reaction). However, with better settlement practices, a bank would be in a position to make more measured adjustments to its FX settlement exposure with a counterparty in response to its evolving financial condition. Thus, if broadly adopted by FX market participants, improved practices for managing FX settlement flows could help stabilise money markets, reduce liquidity pressures and contain the systemic risk in settling, or deciding not to settle, FX transactions at times of market stress.

Reduced uncertainty. If banks were to improve their settlement practices, they would need to eliminate current uncertainties regarding the revocability of their payment instructions and the finality of their receipts. As described in Appendix 1, such uncertainties can stem from correspondent banking arrangements as well as from the rules and laws governing domestic payments. Clearer arrangements, rules and laws would reduce market-wide uncertainty, another source of systemic risk at times of stress.

Impact on monetary policy. Since the envisioned action by individual banks would not require modification of the underlying payments system infrastructure, FX settlements could continue to take place relatively independently in each domestic market during normal business hours. Under such circumstances, private sector sources of liquidity should be as available as they are today to support the settlement of each traded currency. As a result, domestic monetary policy implementation should not be affected by, for instance, a significant change in the demand for central bank balances or for central bank credit and liquidity. In addition, individual central banks would retain their current degree of flexibility in deciding how to respond to liquidity problems in their home currencies.

Increased correspondent banking competition. A market-wide increase in the desire of individual banks to lower their FX settlement exposures should also encourage competition in the quality of correspondent banking services. This demand could stimulate private sector innovation more broadly to develop multi-currency services - including multi-currency settlement mechanisms and bilateral and multilateral obligation netting arrangements - that could help banks achieve even further risk reductions.

Action by non-bank financial institutions. In addition to inducing banks, and their correspondent banks, to make improvements, where appropriate and relevant it would also be necessary to reach non-bank financial institutions active in the FX markets. Unless all relevant market participants took sufficient steps to control their FX settlement exposures, an excessive level of risk might remain in the financial system; moreover, unfair competitive advantages could emerge in the near term if some market participants incurred the cost of improving their FX settlement practices while others did not.

4.2 Action by industry groups to provide risk-reducing multi-currency services

Industry groups are encouraged to develop well-constructed and soundly based multi-currency services that would contribute to the risk reduction efforts of individual banks and would reduce systemic risk more broadly. This recognises the significant potential benefits of multi-currency settlement mechanisms and bilateral and multilateral obligation netting arrangements, and the G-10 central banks' view that such services would best be provided by the private sector rather than the public sector.

4.2.1 Description of recommended action

Payment/receipt relationship. FX settlement exposure could be attacked at its source by creating one or more multi-currency settlement mechanisms that establish a direct relationship between the payment of one currency and the receipt of another. As defined in the Noël Report, a multi-currency delivery-versus-payment (DVP) mechanism would assure participants that "a final transfer in one currency occurs if and only if a final transfer of the other currency or currencies also takes place". Accordingly, a multi-currency DVP settlement mechanism (which this report calls a payment-versus-payment, or PVP, settlement mechanism) would eliminate FX settlement exposure. However, a PVP mechanism would not necessarily reduce or eliminate the liquidity or other risks that can arise when settling FX trades; indeed, some PVP mechanisms might magnify these risks. Moreover, while "PVP" may present a clear theoretical concept, it does not address some of the practical timing problems of coordinated cross-border payments and settlements. Thus, from an analytical perspective, it is helpful to pay particular attention to two important characteristics of a multi-currency settlement mechanism: the payment/receipt relationship and the timing of settlement.

Through discussions with market participants, the CPSS has identified two possible payment/receipt relationships that a multi-currency settlement system could establish. Under one approach, the system could guarantee the timely settlement of all relevant currencies. Such a system would assure participants which fulfil their settlement obligations that they will receive what they are owed even if their counterparties fail. Under another approach, a system would assure participants that if a counterparty fails to meet its settlement obligations then all of their related payments to that counterparty will be returned or cancelled. In the light of these major differences in approach, it is useful to specify these two potential payment/receipt relationships:

Guaranteed receipt system: counterparties are guaranteed that if they fulfil their settlement obligations they will receive on time what they are owed ("you will receive if and only if you pay").

Guaranteed refund system: counterparties are guaranteed that any settlement payment they make will be cancelled or returned if their counterparties fail to pay what they owe ("you will pay if and only if your counterparty pays").

Settling individual trades. Either approach could potentially eliminate FX settlement risk in the settlement of individual FX trades. For instance, a guaranteed receipt system would assure a participant that if it pays the single currency it sold, it will receive the single currency it bought even if its counterparty fails to pay what it owes. In contrast, a guaranteed refund system would assure a participant settling an individual trade that if it pays the single currency it sold but its counterparty fails to settle, then its own payment will be cancelled or returned.

Settling netted trades. In addition, either payment/receipt relationship could be designed to eliminate FX settlement risk in the settlement of FX trades under an obligation netting agreement. For instance, a guaranteed receipt system could assure a participant that if it pays each of the currencies it sold on a net basis under, for example, a bilateral obligation netting agreement, it will receive each of the currencies it bought on a net basis even if its counterparty fails to settle any of its net payment obligations. For the settlement of trades under a multilateral obligation netting agreement, the same guarantee could be offered in the case of a settlement failure by any of the participant's counterparties. In contrast, a guaranteed refund system could assure a participant that all of its payments in all of the currencies it sold on a net basis under a bilateral obligation netting agreement will be cancelled or returned if its counterparty fails to settle any of its payment obligations. For the settlement of trades under a multilateral obligation netting agreement, the system could guarantee the cancellation or return of all payments directed to all defaulting counterparties.

Timing of settlement. Along with its potential payment/receipt relationship, the timing of a particular system's settlement can play an important role. With either payment/receipt relationship, a system could settle all relevant currencies at the same time - simultaneous settlement - or at different times - sequential settlement. With a simultaneous settlement, a system could require all participants to pay in all the funds they owe in every currency before the system pays out any funds that participants are due to receive in any currency. In contrast, in a sequential settlement, a system would pay out some currencies before it receives all other currencies.

Guaranteed receipt with simultaneous settlement. The timing of settlement can have many implications, not the least of which is its impact on the system's ability to honour the guarantee backing its payment/receipt relationship. For instance, a system that provides a guaranteed receipt with simultaneous settlement could benefit from the generally self-collateralising nature of FX settlements. If a participant in such a system were to fail to pay its obligation in one currency, the system could, if well designed, cover this shortfall by using the defaulting participant's corresponding conditional receipts in another currency to purchase or to collateralise the funds needed to pay its counterparties. Of course, exchange rate movements mean that the value of one side of an FX trade will not always equal the other, thereby limiting the self-collateralising capability of a simultaneous settlement. Accordingly, a system that provides a guaranteed receipt with simultaneous settlement would need other sources of collateral or margin to cover this potential shortfall. Such a system could also establish additional risk controls (e.g. limits, committed lines of credit, other sources of liquidity and loss-sharing arrangements) that would give participants further assurance that the system has sufficient resources to honour its settlement guarantee in a variety of situations. The system would also require the appropriate operational controls and legal basis to carry out the settlement as intended.

Guaranteed refund with simultaneous settlement. A system that provides a guaranteed refund with simultaneous settlement could benefit from its ability to verify the irrevocable (although conditional) receipt of all settlement payments before it pays out any funds to its participants. In this way, if any participant failed to meet its settlement obligation in any or all currencies, the system would be in a position to return or cancel any of the corresponding conditional payments from the counterparties to the defaulter. Such a system would also require the appropriate operational controls and legal foundation.

Guaranteed receipt with sequential settlement. If, in contrast, settlement was carried out sequentially, a system would need to find other ways to guarantee its intended payment/receipt relationship. For instance, a system that provides a guaranteed receipt with sequential settlement could establish a set of risk controls that would give it sufficient resources to honour its settlement guarantee in a variety of situations. As in the case of simultaneous settlement, these risk controls might consist of an appropriate combination of limits, committed lines of credit, collateral and loss-sharing arrangements. However, a sequential settlement system would not be "self-collateralising" to the same degree as a simultaneous settlement system since it would be obligated to disburse with finality at least some currencies before it receives others. As a result, the system might need to design different risk controls or to mobilise other, perhaps more costly, sources of collateral to support its guarantee.

Guaranteed refund with sequential settlement. A system that wishes to provide a guaranteed refund with sequential settlement would probably need to rely on some combination of prefunding or delayed disbursement to support its intended payment/receipt relationship. For instance, such a system might require participants to prepay their settlement obligations in late-settling currencies conditionally one day ahead of time. This would permit the system to verify the receipt of all expected funds and, if necessary, to cancel or return on settlement day any payment destined for a counterparty that failed to pay what it owed. However, given the potential funding costs and liquidity requirements that this prefunding process could create, it may be less economical for a system to provide a guaranteed refund with a sequential settlement process than with a simultaneous one.

Payment versus payment. It is worth noting that the CPSS did not find it analytically helpful to use the label "PVP" to differentiate between the four identified settlement mechanisms: guaranteed receipt with simultaneous settlement; guaranteed refund with simultaneous settlement; guaranteed receipt with sequential settlement; and guaranteed refund with sequential settlement. In fact, any of the four systems could arguably be called a PVP mechanism that assures participants that a final transfer in one currency will occur if and only if a final transfer of another currency or currencies also takes place. For instance, in a guaranteed refund system (whether settlement takes place simultaneously or sequentially), each participant's transfer is conditioned upon a related transfer by one or more of its trading counterparties. In contrast, in a guaranteed receipt system (whether settlement takes place simultaneously or sequentially), the transfer to each participant is conditioned upon a related transfer by it to the "system". Whether a particular approach should be labelled "PVP", however, is an insignificant issue compared with its overall effectiveness in eliminating FX settlement exposure and the other risks that can arise when settling FX trades. In practice, this effectiveness will depend critically on the strength and nature of the guarantee supporting the intended payment/receipt relationship of the settlement mechanism.