3.1 Duration of foreign exchange settlement exposures

For most of the banks surveyed and in every G-10 country, the minimum FX settlement exposure of an individual spot or forward trade (the duration of Status I) currently lasts for between one and two business days. In addition, it can take a further one to two business days for many banks to establish whether they indeed received the currency they bought on time (the duration of Status U). As a result, more than three business days - plus any intervening weekends and holidays - can elapse between the beginning of some banks' settlement exposures and the time at which they know with certainty that they are no longer at risk.

Furthermore, banks can incur FX settlement risk no matter which currency they buy or sell. For instance, even when - from one bank's point of view - the bought currency settles before the sold currency, a bank might face an earlier deadline for unilaterally cancelling its payment of the sold currency. In such circumstances, it could be forced to pay out the currency it sold even when it knows that it will fail to receive the currency it bought.

As described in Appendix 1, the often lengthy duration of FX settlement exposure reflects the fact that current practices for handling payments and receipts were designed more for operational efficiency (e.g. faster throughput, lower costs, prevention of technical fails) than for controlling settlement exposures. For instance, many automation advances such as "straight-through processing" have offered worthwhile benefits to payments systems and individual banks by, inter alia, reducing operational risks. Once they begin, however, certain automated procedures can make it difficult, if not impossible, for a bank or its correspondent to cancel unexecuted payment instructions even before settlement day. This can increase the duration of FX settlement exposures by creating overly restrictive unilateral payment cancellation deadlines.

3.2 Size of foreign exchange settlement exposures

Given current practices, many banks face significant FX settlement exposures overnight, and therefore over weekends and holidays. The size of a bank's total FX settlement exposure depends directly on the duration of the settlement exposure of each of its trades. For instance, if a bank's minimum settlement exposure for a single FX trade lasts 48 hours, at least two days' worth of trades would always be at risk. In addition, if it takes, for example, another 24 hours to verify the final receipt of each purchased currency, a further day's worth of trades might still be at risk. Appendix 1 illustrates how, under these circumstances, a bank's maximum FX settlement exposure could equal at least three days' worth of trades at any point in time, including overnight and during weekends and holidays.

No comprehensive statistics are yet available on the banks' actual levels of FX settlement exposure, partly reflecting the fact that most banks currently do not measure them properly, if at all. Nevertheless, discussions with several banks indicated that their current exposures can reach very high amounts. For instance, some banks said that they routinely settle FX trades worth well over US$ 1 billion with a single counterparty on a single day. If current practices can transform this level of activity into an actual FX settlement exposure that is two to three times this amount, bilateral FX settlement exposures could be large in relation to a bank's capital and could far exceed the short-term credit exposure a bank incurs in other activities with the same counterparties.

3.3 Potential impact of changing practices

The survey indicated that a bank's FX settlement practices can greatly influence the size of its exposures. One way a bank can lower its exposure is by changing the timing of its unilateral payment cancellation deadlines and of its identification of final and failed receipts. Another way is by legally binding netting of the daily settlement obligations arising out of its FX trades rather than settling each trade individually.

3.3.1 Payment cancellation and receipt identification

A bank could eliminate overly restrictive unilateral payment cancellation deadlines (to shorten the duration of Status I) and reduce the time it takes to identify its final and failed receipts of bought currencies (to shorten the duration of Status U). As described in Appendix 1, these improvements could require a combination of changes to its own settlement practices and, if relevant, to its correspondent banking arrangements. Some banks have also proposed that the establishment of a global practice to attach a common reference number specific to each FX trade and its related payment instructions could be quite helpful in this regard.

In October 1994 the New York Foreign Exchange Committee (NYFEC), which is a private sector group sponsored by the Federal Reserve Bank of New York, published a report on Reducing Foreign Exchange Settlement Risk. In its report, the NYFEC defined "best-case" FX settlement practices as those that would give a bank the following capabilities:

• To cancel its payment instructions unilaterally up until the opening time on settlement day of the local large-value transfer system (LVTS)
• To identify its final and failed receipts immediately upon finality of the local LVTS

While there may be different views as to what constitutes "best practice" in different markets, the NYFEC's definition provides a useful reference point for measuring the effect of changing current settlement practices. For instance, Appendix 1 illustrates how current "worst-case" market practices (i.e. the earliest cancellation deadlines and the latest receipt identification times reported in the market survey) can produce FX settlement exposures that are two to three times greater than those that would be generated by the NYFEC's "best-case" practices. This is particularly noteworthy since the NYFEC's "best-case" practices are already being followed by at least some market participants in different G-10 countries, providing concrete evidence that similar practices could be adopted immediately by all participants in the FX market.

3.3.2 Netting

There is also scope for at least some FX market participants to reduce the amount at risk during the settlement process through obligation netting arrangements. As defined in this report, obligation netting is the legally binding netting of amounts due in the same currency for settlement on the same day under two or more trades. Under an obligation netting agreement for FX transactions, counterparties are required to settle on the due date all of the trades included in the agreement by either making or receiving a single payment in each of the relevant currencies. This reduces the amount at risk by lowering the number and size of payments that would otherwise be needed to settle the underlying transactions on a trade-by-trade basis.

Appendix 1 illustrates the potential exposure-reducing benefits of obligation netting. Most importantly, any actual reduction in FX settlement exposures would depend on a bank's trading pattern. Active market-makers trading with each other out of a limited number of locations would be likely to have many offsetting trades that could be netted, whereas relatively inactive traders, or those that trade out of many different locations around the world, might have less opportunity to net their FX trades.

3.4 Market responses and initiatives

3.4.1 Individual bank level

Risk awareness. Although most banks are familiar with the concept of foreign exchange settlement risk, not all banks have a single officer who understands the entire settlement process and the risks it entails, so that meetings with some banks for the purpose of the market survey required the presence of representatives from several different departments in order to cover all aspects of the subject. Moreover, some market participants indicated that the senior executives of their banks had never been fully briefed on the FX settlement process and the associated risks. Some of the bankers interviewed saw this CPSS project as an effective way to alert senior management to the risks posed by a bank's FX settlement process and to obtain a clear mandate to improve practices.

Many bankers also suggested that it would be helpful to publicise information on the size of actual losses that banks have incurred in the past during the settlement of FX trades with problem counterparties. Such data, however, are not readily available across markets, in large part owing to banks' reluctance to make these figures publicly available. Some market groups are attempting to estimate the industry-wide cost of current FX settlement exposures and to prorate these costs to the individual bank level. While such estimates would, of necessity, be based on many assumptions and approximations, they might prove to be a useful way to highlight the potential mismatch between the risks and rewards that many banks currently face when settling their FX trades.

Risk measurement. Overall, many banks currently underestimate the duration and size of FX settlement exposure by treating it as an intraday amount no larger than a single day's expected receipts. Only a few banks treat irrevocable payment instructions issued prior to settlement day as part of their FX settlement exposure. In addition, most banks do not appear to incorporate due but unverified receipts, let alone failed receipts, in their measures of outstanding exposure with a counterparty. As a result, many banks do not recognise that they can routinely incur FX settlement exposures equivalent to several days' trades, and that these exposures can persist overnight, and therefore over weekends and holidays.

Systems for measuring FX settlement exposures vary widely. Some banks have no formal mechanism to measure their current or future settlement exposures, while others project their settlement exposures with direct information feeds from their trading systems. Some of the latter banks project exposures up to three days ahead, while others only look at the next day's amount.

Few banks, however, appear to draw together in a useful way all the relevant information potentially available to them that could be used to accurately measure their exposure throughout the settlement process. Hitherto, economic incentives have not been sufficiently strong for banks to assemble this existing information from their back offices and correspondent banks in a useful fashion. For instance, cash managers may become aware of a failed receipt on settlement day, but it is unclear how easily they can determine which counterparty failed to make which payment. In some cases a bank will send its correspondent an "advice to receive", which is a notice of a payment the bank expects from a third party. Some correspondents compare these advices with actual receipts and alert the beneficiaries to failed receipts. However, banks generally do not use this practice. Some correspondents charge heavily for this service; others either actively discourage their customer banks from sending such advices or ignore them if they are sent. In some cases, of course, advice from a correspondent to a cash manager of a failed receipt, or of an unexpected overdraft, will arrive too late for the manager to take any action that day - the manager may not even be able to tell, until the next business day, which anticipated receipt has failed, from which counterparty.

Most of the banks surveyed agreed on a general, theoretical level with the CPSS's methodology for defining and measuring FX settlement exposures. Only a few banks, however, expressed interest in implementing an internal measurement system that would be fully consistent with this framework. In large part, this reflects concern about the cost of such a system, particularly one that would continually update a bank's global exposures as it executes each new trade and as each unsettled trade moves through the settlement process. It also reflects different views on the level of detail that a bank would need in practice to control its FX settlement exposures. For instance, some banks believe that they would only need to measure their minimum exposure (i.e. the sum of their Status I and F trades), while others believe that they would only need to measure their maximum exposure (i.e. the sum of their Status I, F and U trades). Some banks believe that failed receipts (Status F trades) should be excluded from their day-to-day measure of FX settlement exposure, while others believe that they would need to include all unsettled trades, including those that have not yet passed a bank's unilateral cancellation deadline (Status R trades).

Risk controls. Some banks currently impose no limits on their FX settlement exposures, no matter how they are measured. As for those banks that do have limits, some use them as effective controls while others set them at extremely high levels or waive them altogether for their largest trading partners. In addition, some banks impose binding settlement limits, particularly on certain counterparties or in special circumstances, while others use such limits only as guidelines. In general, banks with FX settlement limits tend to set them with an eye to preventing unusual trading activity rather than containing credit and liquidity exposure. As a result, limits and actual FX settlement exposures can reach multiples of those encountered in the case of products with similar risks such as overnight placements and deposits involving the same counterparties.

Some banks are considering the introduction of new settlement limit systems. However, some believe that even if they were to properly measure and project their exposures, it would be difficult for them to introduce binding counterparty settlement limits if others in the market did not do the same thing. In addition, some banks said that even if they had an effective limit system, market pressures might make it difficult for them to cut their exposures to certain counterparties quickly in response to increased credit concerns.

Prospects for individual action. Many banks have not clearly established responsibility for managing foreign exchange settlement risk within their institutions, and so have not created the authority and incentives to control it prudently. While some bankers plan to spend the necessary time and money to improve their settlement practices, others do not and see little incentive to change their practices without a strong mandate from their senior management. This resistance appears to stem from a combination of sources, including:

• Failure to recognise that banks can routinely incur significant FX settlement exposures overnight and during weekends and holidays
• A mistaken view that an FX settlement exposure with a counterparty represents less of a risk than a loan or other formal credit extension of the same size and duration
• False comfort that major FX counterparties are "too big to fail"
• A misperception that public authorities could always avoid closing down a major FX market participant unexpectedly at a time that would cause significant losses during the settlement process
• The current complexity of settling FX transactions
• Concern that the necessary improvements will be very costly
• Fear that an uneven pace of improvements among individual market participants could lead to competitive distortions
• Other, higher priorities

3.4.2 Industry group level

Risk awareness. Much promising work has begun to take place at the industry group level. As mentioned above, the New York Foreign Exchange Committee (NYFEC) published a report in 1994 on Reducing Foreign Exchange Settlement Risk. This study documented for the first time the considerable impact of market practices on the size and duration of FX settlement exposure. This finding, which the CPSS subsequently confirmed through its survey of banks throughout the G-10 countries, led the NYFEC to issue a set of recommendations designed to help banks immediately reduce their FX settlement exposures (see Appendix 2).

At the international level, the NYFEC publicised the report through seminars in New York, London, Frankfurt and Tokyo. The Association Cambiste Internationale (ACI) has also expressed a strong interest in supporting efforts to encourage improvements in FX settlement practices. In general, the topic of FX settlement exposure has received considerable attention at many recent industry conferences, including the 1995 S.W.I.F.T. International Banking Operations Seminar (SIBOS).

Bilateral netting services. FXNET, S.W.I.F.T. and VALUNET currently provide bilateral obligation netting services to many banks. As of December 1995, FXNET provided this service to 29 institutions operating out of 57 offices in 9 locations, including New York, London, Zurich, Tokyo and Singapore, and an additional 19 offices were in the process of joining the system. Three new locations (Geneva, Sydney and Toronto) would be introduced with the planned expansion. Accord, which is operated by S.W.I.F.T. (Society for Worldwide Interbank Financial Telecommunication), provides confirmation matching and bilateral obligation netting services. As of December 1995, 370 users employed the Accord matching services, including 27 subscribers to its netting services. VALUNET, the smallest of the service providers, is operated by International Clearing Systems (service provider for the proposed Multinet International Bank, which is discussed below). As of December 1995, VALUNET provided bilateral obligation netting services to 10 institutions operating out of 17 offices in 5 locations. In addition to these industry services, many pairs of banks have set up bilateral netting arrangements on their own, often using a standardised contract such as the International Foreign Exchange Master Agreement (IFEMA).

However, despite the potential risk-reducing benefits, the market survey indicated that not all banks use bilateral obligation netting agreements. When they do net, more often than not their netting is limited to close-out provisions (mainly to take advantage of favourable capital treatment of netted positions or to improve their leverage ratios), while routine settlements continue to be conducted on a gross, trade-by-trade basis. Obligation netting is mostly confined to the largest banks and their largest counterparties.

Banks cite costs and operational capacity as barriers to the greater use of bilateral netting by novation or other methods of obligation netting. In several countries, banks also expressed concern about the lack of legal certainty of netting arrangements. Some of the discussions suggested a possible role for the European Commission or for some individual central banks in validating netting contracts. In contrast, some banks find it cost-effective to informally settle their foreign exchange trades by paying and receiving their obligations on a net basis. However, uncertainty regarding the legal soundness of such arrangements could potentially increase systemic risk.

Multilateral netting and settlement services. ECHO (Exchange Clearing House) began operations in August 1995 and the proposed Multinet International Bank hopes to start in 1996. Both systems are designed to transform bilaterally arranged individual FX trades into multilateral net settlement obligations and to provide risk controls that ensure the timely settlement of these obligations. In essence, these controls are designed to reduce credit and liquidity risks by assuring participants that the final settlement of each currency will take place even if a participant in the group is itself unable to settle its obligations on the due day.

ECHO began operations with 16 participant users in 8 countries netting trades in 11 currencies. ECHO hopes to expand its services to banks in more than 20 countries for trades in 25 currencies. Multinet plans initially to provide services to 8 banks in North America for their trades in US and Canadian dollars and other major currencies within the first year. Multinet also hopes to add further currencies and participants in other countries over time. It may be noted that central banks have successfully used the minimum standards and cooperative oversight principles set out in the Lamfalussy Report when reviewing these systems.

Other multi-currency settlement mechanisms. More recently, the newly formed "Group of 20" has been actively exploring other possible multi-currency settlement mechanisms. Rather than directly netting the underlying FX trades, the models currently under study could be designed to support the settlement of individual trades or trades which have already been netted under other bilateral or multilateral obligation netting arrangements. Although these multi-currency settlement mechanisms would not, by themselves, provide the risk-reducing benefits of obligation netting, they could lower credit risks by assuring participants that the final transfer of one currency will occur if and only if the final transfer of the other relevant currency or currencies also occurs. Multi-currency settlement mechanisms could also, if designed accordingly, lower liquidity risks by assuring participants that if they settle their payment obligations then they will receive their expected funds on time.

Prospects for collective action. Although some of these industry-wide initiatives are well under way, many banks remain sceptical about the business case for committing resources to efforts to reduce FX settlement exposures. As a result, many individual banks have been slow to join these efforts. Without adequate motivation for a sufficient number of FX market participants to support and use one or more of these current or prospective industry-wide multi-currency services, their short-term (let alone long-term) viability is uncertain.