2.2 But netting can also contribute to an increase in systemic risk. This may be the case if, instead of achieving reductions in participants' true exposures, it merely obscures the level of exposures. Effective reductions in actual exposures depend on the legal soundness of a netting scheme. But even when actual exposures are reduced, multilateral netting systems may shift and concentrate risks in ways that will increase systemic risk by increasing the likelihood that one institution's failure will undermine the financial condition of others. Containing this aspect of systemic risk depends on both the netting provider and the participants being aware of the sources and extent of their exposures, on the strength of their incentives to contain these exposures and, ultimately, on their ability to absorb losses in the event of a default.

Bilateral netting.

2.3 Bilateral netting has been applied recently to spot and forward foreign exchange and swap contracts. In general, it can reduce bilateral payment flows to net amounts and, thereby, reduce operational costs and liquidity needs. By reducing the reciprocal amounts to be settled on each value date for foreign exchange contracts, for example, bilateral netting can also reduce the size of the cross-currency settlement risks which a bank incurs when it pays out funds in one currency before receiving payment in another. Finally, bilateral netting arrangements can lower a bank's counterparty credit exposure by permitting netting of unrealised losses against unrealised gains on outstanding forward contracts in the event of a counterparty's closure.

2.4 "Netting-by-novation" agreements provide for individual forward-value contractual commitments, typically foreign exchange contracts, to be discharged at the time of their confirmation and replaced by new obligations forming part of a single agreement. Amounts due under a discharged contract will be added to running balances due between the parties in each currency at each future value date. This reduces the size of the payments to be made on each value date and their number to one per currency traded. In some markets participants may be able to achieve a reduction of more than 50% in total payments in all currencies, both in terms of value and volume. Routine liquidity needs and operational costs and risks are thereby reduced. By merging all of the transactions into a single agreement, covering the running accounts, the credit exposure that would be faced in respect of forward commitments in the event of a counterparty default can be reduced to a single net amount. This could be expressed as the discounted present value of the sum of the reciprocal future payments called for in the running accounts, and could also be expressed as the discounted present value of the net of unrealised gains and losses on all of the obligations included in the netting contract. In some countries, similar results can be obtained through the use of a "current account" agreement.

2.5 "Master agreements", which have been principally applied to foreign exchange and interest rate swap contracts, also permit the incorporation of a number of individual contracts between two parties into one legal agreement and, thereby, bring about a similar, single net-credit exposure with respect to forward commitments. In contrast to what occurs in the case of novation, the individual transactions are not "blended" into running accounts but retain their specific terms, rates and maturities so that they can be individually assigned or terminated. Master agreements can include provisions for the netting of payments on value dates.

2.6 In some jurisdictions, a liquidating authority of a failed financial institution, in seeking to maximise the assets of the estate of the closed institution available to its general creditors, can engage in "cherry picking" by performing only those forward contracts that are profitable to the estate while repudiating those contracts that are unprofitable. The contracts that have become unprofitable to the estate are precisely those that are profitable to the closed institution's counterparty. If the authority were successful in cherry picking, the counterparty's credit exposure could be the sum of the "gross" exposures on its profitable contracts.

2.7 But if the counterparty has a legally enforceable netting agreement, the liquidating authority is obligated either to perform all of the transactions included (or to undertake payment of all of the amounts owed on the running accounts) or to default on the entire agreement. The counterparty's credit exposure in event of default is not the gross exposure on its profitable contracts but the "net" cost of replacing all the obligations (or payments) due under the agreement - summed across currencies and value dates - an amount equivalent to the balance of unrealised gains and losses on all forward commitments (or all payments made and received).

Multilateral netting systems.

2.8 Multilateral interbank netting systems have so far been principally applied to payment orders. These systems typically represent an extension of the traditional correspondent banking service of making book transfers. However, commercial banks are actively developing proposals to establish organised multilateral netting systems for spot and forward foreign exchange contracts now traded in the "over-the-counter" interbank markets. Although their proposals are not yet fully developed, three groups of bankers - in Canada, Europe, and the I~7nited States have made considerable progress in the last year and they have begun to share information.

2.9 Netting on a multilateral basis is arithmetically achieved by taking the sum of each participant's bilateral net positions with each of the other participants. This produces a single multilateral "net-net" position. When the netting is conducted through a central entity which is legally substituted for the original counterparty to each payment or contract, the net-net position will, in fact, constitute a bilateral net position between each participant and the central counterparty.

2.10 In payment netting systems participants with net-debit positions will be obligated to make a net-settlement payment to the central counterparty who, in turn, is obligated to pay those participants with net-credit positions. The net-credit and net-debit positions of all participants sum to zero. Thus, if any participant is unable to meet its net-debit position, the central counterparty incurs a corresponding short-fall.

2.11 In forward contract netting systems the relationship of the central counterparty to each participant would be similar in many respects to that established under bilateral netting-by-novation agreements. Once a contract entered into by two participants is confirmed and accepted by the central counterparty, it would be discharged and the amounts due under it would be added to running accounts between the central counterparty and each participant. In the event of a participant's default, the central counterparty would face a forward credit exposure equal to the replacement cost of the entire stream of payments due under the accounts. It might also face liquidity and cross-currency settlement exposures at the time of the default.

2.12 Both forms of multilateral netting can bring about substantial reductions in payment flows and liquidity needs and, if they are legally binding, in the level of credit and liquidity exposures faced by the participants. Although the effects depend upon settlement and trading patterns, private payment netting systems can evidently reduce the value of settlement payments and the number of payments by 80% or more from what would be needed for gross settlements in certain cases. Multilateral contract netting systems may be able to achieve similar results with respect to the value and volume of settlement payments and, at the same time, reduce the level of forward credit exposure on included foreign exchange transactions to a comparable extent.

2.13 It should be noted that these reductions relate only to the aggregate liquidity demands and credit losses to be covered by the central counterparty or shared in some way among the surviving participants, in the event of a participant's default, compared to the sum of individual exposures that would have occurred on the same set of transactions in the absence of netting. Whether an individual surviving participant would experience smaller liquidity demands or credit losses on the same set of transactions would depend on the structure of the netting system and its particular loss-allocation procedures.

2.14 For example, if loss-sharing rules mutualise the losses from any one participant's failure among all surviving participants, either on the basis of equal shares or of overall trading volumes, then participants are exposed to losses caused by the failure of other participants with whom they may have had only limited bilateral dealings or even none at all. This reflects a shifting of a portion of the risks from those participants that traded with a defaulting party to others who did not.

2.15 Moreover, as discussed below in Section 3, if use of multilateral netting induces or permits participants to expand their trading activities significantly, net exposures conceivably could return to, or even exceed, the level of (gross) exposures that existed in the absence of netting.

Cross-currency settlement risks.

2.16 By reducing the reciprocal amounts due on a set of foreign exchange contracts, netting can reduce the size of the cross-currency settlement exposures entailed in separate settlements of the different currencies. According to some market participants, "Herstatt" risk - named for the 1974 failure of Bankhaus Herstatt - accounts for more than half of the credit exposure incurred in their foreign exchange trading. The desire to reduce these exposures provides banks with a strong incentive to adopt netting. But while the size of these exposures can be reduced, neither bilateral nor multilateral netting arrangements can eliminate cross-currency settlement risk entirely.

2.17 The duration of Herstatt exposures is particularly long in the case of currency pairs for which there is no overlap between the hours of operation of the central banks of issue or their respective domestic payment systems. However, cross-currency settlement exposures are also faced in the settlement of any currency pairs in the absence of organised mechanisms for simultaneous final settlement of the different currencies. While a multilateral foreign exchange contract netting system could greatly reduce the amounts to be settled, it would concentrate cross-currency settlement risk on the central counterparty, which would have to manage Herstatt exposures to each of the system's participants.

Third-party effects.

2.18 Netting can be perceived to operate to the disadvantage of the unsecured creditors of a failed participant relative to those of other netting participants. This view is based on an ex post analysis from the perspective of the shareholders, other investors and depositors of a participant that has been closed. But in an ex ante analysis of the credit risks incurred by all unsecured creditors of netting participants, before the fact of any one institution's insolvency, the reduction in exposures experienced by all netting participants will benefit all of their unsecured creditors - even if, in the event of one participant's closure, it also benefits some at the expense of others. Moreover, it can be argued that in the absence of netting the unsecured creditors of surviving participants tend to be disadvantaged relative to those of a failed institution.

2.19 By preventing a liquidating authority from cherry picking among a closed institution's profitable and unprofitable forward contracts, legally enforceable bilateral netting reduces the magnitude of the credit losses experienced by the institution's netting counterparties. There is a zero-sum relationship between this reduction in credit losses for netting participants and the increased risk of loss to their other unsecured creditors. But it appears that in the majority of the G-10 countries a bank should be able to assert and successfully defend so me form of statutory or common- law right of set-off that would produce a similar net exposure even in the absence of a written agreement. In these countries bilateral netting agreements provide added certainty of the result of net exposures, in the event of a counterparty's closure, and may also provide for the routine netting of payment flows between counterparties; other creditors will be no worse off.

2.20 The significant reductions in credit exposures produced for participants in multilateral netting systems go a step further. To the extent that there are reductions in the size of the aggregate loss to be shared by the surviving participants - by offsetting what would be the net claims of the estate of the defaulting participant on some participants against other participants' net claims on the estate - multilateral netting necessarily reduces the assets of the estate of the closed participant and, thus, increases the losses to be shared by its other unsecured creditors. This adverse effect could be exacerbated to the extent that obligations to the central counterparty are collateralised with high-quality assets of the closed participant.

2.21 But by reducing credit and liquidity exposures experienced in trading and dealing activities, legally binding bilateral and multilateral netting should benefit netting participants and all of their unsecured creditors in several ways. As mentioned above, on an ex ante basis netting should benefit the whole class of unsecured creditors of participants even if, in the event of an insolvency, it disadvantages some relative to others. Moreover, the reductions in exposures relative to capital should, ceteris paribus, reduce the likelihood that disturbances outside the netting scheme will lead to any one participant's insolvency. The reductions in exposures should also reduce the risks of secondary defaults within the group of netting participants and, thus, contribute to a reduction in systemic risk.

2.22 Moreover, if the combined effect of the ongoing reduction in the size of exposures together with the limitation on the liquidator's ability to cherry pick, as well as any use of collateral, increases overall credit risks to some unsecured creditors then, over time, at least institutional investors should be expected to demand compensation in the form of higher returns. This would depend, however, on the availability of information on both the extent of netting and on the use of collateral, as well as on the willingness of banks to engage in differential pricing.

The Importance Of Legal Enforceability

2.23 The Committee's analysis confirms the view, expressed in the Angell Report, that netting schemes only reduce credit and liquidity exposures if they have a sound legal basis. Any netting scheme which provides for the settlement of net balances, on large numbers of transactions, can reduce the participants' day-to-day operating costs and routine liquidity needs. But only if the net amounts are legally binding in the event of a counterparty's closure will the participants experience reductions in their true credit and liquidity exposures.

2.24 The concept of netting, in its broadest sense, is given effect under the law of each of the G-10 countries. Bilateral master agreements and either bilateral netting-by-novation or current account agreements are likely to be enforceable in all eleven countries. Thus, where a properly-prepared written netting agreement has been established, the surviving party's credit exposure in the event of a counterparty's closure would be the balance of unrealised gains and losses on all included transactions. (As indicated above, in the majority of G-10 countries the same net-credit exposures are likely to occur even in the absence of a written netting agreement.)

2.25 Multilateral payment netting through a central counterparty is likely to produce legally binding net positions (i.e., "net-net" positions) against the central counterparty, in the event of a participant's closure, in the majority of the eleven countries. This result, however, is unlikely in those countries where the expected application of bankruptcy law retroactively renders transactions of a closed institution ineffective after 0.00 a.m. on the date it is ordered closed. However, in all eleven countries multilateral netting of forward foreign exchange contracts through a central counterparty is likely to be legally enforceable in the event of a participant's closure, with the exception only of those contracts entered into on the date of the closure in those countries with a "zero-hour" bankruptcy rule.

2.26 In each country various conditions must be met for the net amounts to be binding, firstly, between the parties and, secondly, upon the liquidating authority of a closed counterparty. In no country should it be assumed that exposures will be the balance of unsettled payment orders made and received or of unrealised gains and losses unless the participants take some steps to try to make them so. Moreover, binding net exposures may not be achievable by all banks in all factual circumstances. Cross-border netting arrangements, for example, between banks chartered or located in different countries, raise choice-of-law and conflict-of-law questions that cannot be easily resolved. Indeed, a complete resolution of these remaining uncertainties may only be possible through the harmonisation of the relevant national laws. Thus, notwithstanding the foregoing general conclusions on the likely legal effectiveness of netting arrangements , establishing a sound basis for the assertion of net exposures will require thorough legal preparation by the participants and netting scheme providers and may not be possible in all instances.

2.27 Should financial institutions come to rely for the purposes of setting dealing and other credit limits, or for managing their liquidity, on exposures produced by netting arrangements that are not legally enforceable, the participants will face credit and liquidity exposures significantly -possibly many times - larger than they expected in the event of a counterparty default. Thus, netting schemes that are not legally effective (or have no legal basis) and which induce participants to rely on net figures while exposures remain at gross levels will serve to increase the risk of secondary defaults.

Systemic Risks In Multilateral Netting Systems

2.28 Even when legally effective in producing net exposures, multilateral netting systems also have the potential to increase systemic risks because they concentrate risks on the central counterparty. If a system is able to manage and contain prudently its exposures to the participants, and to the extent that in the event of one participant's default it is able to continue to satisfy its obligations to the others, then the system's operation will contribute to a reduction in the level of exposures experienced by participants and also to a reduction in the level of systemic risk. If these conditions are not met the system's operation will increase the likelihood that the credit or liquidity problems of one market participant will suddenly and negatively affect the condition of others. Moreover, the concentration of risks on the central counterparty exposes all participants to the risk that the central counterparty itself may fail.

2.29 The Angell Report concluded that realisation of the overall risk-reduction potential of multilateral netting critically depends on the "financial condition" of the central counterparty. This reflected an assumption that all risks would be borne and managed by the central counterparty, as is the case in systems with "centralised" risk-management procedures similar to those employed by the clearing organisations associated with options and futures exchanges. In such arrangements, the central counterparty typically requires participants to post collateral or margin to secure fully the system's exposure to each of them. Collateral requirements also serve to place limits on the level of participants' activities. In centralised payment netting systems, a bank providing the netting service may not require all participants to post collateral and may regard open exposures as a part of routine correspondent credit facilities. In either case, the participants are not able to contain exposures to individual counterparties and have no incentive to do so. As a result, the ability of the system as a whole to withstand the default of individual participants, or other adverse developments, depends entirely on the risk-management procedures of the central counterparty and its financial ability to absorb losses.

2.30 Both payment and contract netting systems can employ an alternative, "decentralised" approach in which participants retain significant responsibilities for risk management. Multilateral netting against the central counterparty would be used to produce legally binding net positions. But in the event of a participant's default, credit losses associated with its net position vis-à-vis the central counterparty would then be allocated on a pro-rata basis among the surviving participants based on their bilateral dealings with the defaulter. Participants would have an incentive to set bilateral limits for their exposures to other participants and these could constitute the principal risk-control mechanism.

2.31 The viability of this type of system, not specifically addressed in the Angell Report, depends less on the ''financial condition" of the central counterparty itself than on the ability of the participants to manage and satisfy their "contingent obligations" to the central counterparty under the loss-sharing formula. Even so, a degree of centralised risk management by the central counterparty would be necessary to ensure that the sum of the bilateral credit exposures, permitted under a participant's counterparty limits, was commensurate with its financial resources.

2.32 In principle both the centralised and decentralised approaches to risk management can be designed to include credit and liquidity safeguards that should ensure the system's ability to manage its exposures and complete daily settlements. The centralised approach relies on the central counterparty alone to manage all exposures and either on its own assets or on collateral posted by participants to ensure the system's ability to satisfy its obligations. The decentralised approach relies principally on a combination of loss-sharing rules and self-administered exposure limits to give participants the incentives and capabilities to manage their own exposures and, therefore, relies ultimately on the participants to satisfy their contingent obligations in the event of a crisis. Of course, in practice, most systems are likely to incorporate some elements of both models.

2.33 In general terms, however, the centralised, collateral-based approach does appear likely to provide somewhat greater protection against systemic risk because the very presence of the participants' collateral or other assets will help to ensure the system's ability to absorb credit losses and manage its liquidity needs. But this is at the cost of the use of the necessary collateral. A purely decentralised approach, on the other hand, would avoid the costs of collateral and provide greater incentives for participants to manage their own exposures.

2.34 While a decentralised approach to the allocation and management of credit exposures is certainly feasible, it is not certain that a completely decentralised approach to the management of a netting system's liquidity risk (which allocated pro-rata shares of a shortfall on the settlement date) could be prudently implemented in the multi-currency environment in which a multilateral foreign exchange contract netting system would operate. Nevertheless, decentralised risk-management systems could be constructed which would include collateral facilities to guarantee the performance of the participants' contingent obligations. Such systems could combine the beneficial incentive effects of decentralised risk management while maintaining the necessary assurance of settlement.