Because principal risk is the single largest source of risk in the settlement process, the Report concluded that the achievement of DVP is critical. However, even if DVP is achieved, the Report cautioned, other risks exist that may be potential sources of systemic problems. For example, no securities settlement system eliminates replacement cost risk, that is, the risk of the loss of an unrealised gain on an unsettled trade because of default by the counterparty prior to settlement. This risk can be reduced by compressing the settlement interval, that is, the amount of time between the execution of a trade and its settlement, but it could be eliminated only by settling trades in real time as they are executed, a practical impossibility given current technology and institutional arrangements.

Even more important from a systemic risk standpoint, DVP does not eliminate liquidity risk, that is, the risk that a counterparty will not settle an obligation when due, but on some unspecified date thereafter. Settlement systems do not eliminate fails. They often provide for the lending of funds or securities to participants to facilitate settlements and reduce liquidity risks, but the amount of credit available is typically limited. While liquidity problems from failures to settle occur day in and day out and are ordinarily quite manageable, they have the potential to create systemic problems if they occur in an unsettled financial environment, for example following a market break or during a recessionary period. In such an environment, failures to settle when due may undermine confidence in the creditworthiness of counterparties, inducing some participants to withhold deliveries or payments and, in turn, preventing others from meeting their obligations.

Finally, DVP cannot eliminate cash deposit risk, that is, the credit risk associated with the holding of cash balances with an intermediary for the purpose of settling securities transactions. This risk can be eliminated by the use of central bank money for payments, but in many countries non-residents and domestic non-bank participants in the securities markets do not have access to central bank funds accounts. Consequently, some or all participants in a settlement system must make payments through funds accounts at commercial banks and cannot avoid cash deposit risks. Nonetheless, cash deposit risks can be substantially mitigated if the money balances used to make payments are same-day funds, that is, if the funds can be transferred or withdrawn on the day of receipt.

With respect to ways of achieving DVP (or, perhaps more accurately, of linking a funds transfer system to a securities transfer system (a CSD)), the Report's review of arrangements in the G-10 countries revealed a wide variety of approaches. Although not all of these arrangements could be fitted neatly into any simple taxonomy, the Report distinguished three broad structural approaches:

Model 1:
systems that settle transfer instructions for both securities and funds on a trade-by-trade (gross) basis, with final (unconditional) transfer of securities from the seller to the buyer (delivery) occurring at the same time as final transfer of funds from the buyer to the seller (payment);

Model 2:
systems that settle securities transfer instructions on a gross basis, with final transfer of securities from the seller to the buyer (delivery) occurring throughout the processing cycle, but settle funds transfer instructions on a net basis, with final transfer of funds from the buyer to the seller (payment) occurring at the end of the processing cycle;

Model 3:
systems that settle transfer instructions for both securities and funds on a net basis, with final transfers of both securities and funds occurring at the end of the processing cycle.

Consequently, the Report concluded that, regardless of the structural approach taken to achieving DVP, the key issue to be addressed in assessing a system's implications for systemic risk is how well it could cope with the failure of one or more participants (or their guarantors, if any) to repay such credit extensions. The strength of a system depends critically on the safeguards employed by the system operator to limit potential losses and liquidity pressures from such a failure. The Report's review of systems in the G-10 countries revealed that the safeguards employed varied considerably from system to system. All of the systems that were reviewed employed membership standards for participants, but the use of other risk controls was far from uniform. Some systems limited credit risks by attempting to ensure that credit extensions were collateralised, either by securities received by the participant during the processing cycle or by other securities. But very few systems imposed a binding collateral requirement, in the sense that completion of a securities transfer to a participant was conditional on the availability of collateral with value greater than or equal to the resulting overdraft or funds debit balance. Likewise, only a few systems imposed binding limits (caps) on overdrafts or funds debit balances, and conditioned completion of a securities transfer on the recipient's money settlement obligation remaining beneath its cap.

Even if such stringent controls are in place, however, systems should establish a clear understanding as to how losses and liquidity pressures from a participant's failure to repay a credit extension would be distributed. Even a binding collateral requirement cannot eliminate such losses or pressures. It is always possible, and in a market collapse it is quite likely, that the liquidation value of the collateral will fall short of the amount of credit extended. Moreover, the value of the collateral would usually not be realisable sufficiently quickly to allow the system to meet obligations to participants in a timely fashion. For example, a sale of the collateral may not be settled for several days. For collateral to be useful in meeting liquidity pressures, it may therefore be necessary to have in place a commitment from one or more banks to lend against the collateral in specified amounts or at specified discounts.

As far as they go, the analysis and conclusions in the DVP Report are equally applicable to cross-border settlements. In fact, the only inherent differences between risks in domestic settlements and in cross-border settlements are differences in legal risks and the potential for foreign exchange settlement risks to arise in a cross-border context. By definition, in a cross-border settlement at least one of the counterparties to the trade is located outside the country in which settlement takes place. Consequently, as discussed further in Annex 3, legal risks that arise in a cross-border settlement may be affected by laws in the country in which the non-resident counterparty is located and, therefore, may differ from legal risks in settlements between resident counterparties in the country in which settlement takes place. In particular, the assessment of legal risks in cross-border settlements is often complicated by choice of law and conflict of laws issues. Choice of law issues relate to ambiguities as to which law most appropriately governs the relationship between the parties involved. Conflict of laws issues arise when the laws of two or more countries that apply to a transaction require different results.

The other inherent difference stems from the need for non-resident counterparties to effect money settlements in foreign currencies. A non-resident usually needs banking facilities in the country of issue of the currency used in the settlement, and it may be able to meet a substantial part of its liquidity needs in the local market. Nonetheless, a non-resident may often need to supplement its liquidity in the local currency by drawing on liquidity in its home country and converting the proceeds into the local currency through foreign exchange transactions. As discussed in a recent report by the Committee, the settlement of foreign exchange transactions involves replacement cost risks and liquidity risks similar to those that exist in securities settlements. In addition, because of differences in the hours of operation of national payment systems, DVP is rarely achieved in foreign exchange settlements. Consequently, counterparties to foreign exchange transactions typically face principal risks, which are commonly termed Herstatt risks in reference to the losses of principal suffered by counterparties of Bankhaus Herstatt following its failure in 1974.

More important than these inherent differences, however, is the fact that the analysis presented in the DVP Report is limited in certain respects, and those limitations are generally far more significant in a cross-border context than in a domestic context. By far the most important limitation is that the DVP Report focused heavily on the settlement by individual CSDs of trades between their direct participants. Even in domestic settlement systems, many buyers and sellers of securities are not direct participants in the CSD. Participation is typically restricted to banks and broker-dealers, and many banks and broker-dealers choose not to participate directly. Instead, they hold their securities and settle their trades through custodian banks (local agents). But, as discussed in the previous section, in a cross-border context the use of local agents or other intermediaries for holding securities and effecting settlements is pervasive, and the DVP Report's focus on direct participants is a critical limitation.

The involvement of other intermediaries in the settlement process heightens the importance of several issues that received little or no attention in the DVP Report. Perhaps the most basic issue is that when a non-resident (or any other party) holds its securities through an intermediary, it is exposed to custody risk - the potential loss of the securities held in custody in the event that the intermediary becomes insolvent, acts negligently or commits fraud. The degree of custody risk is influenced by a variety of factors. These include the legal status of the securities, the accounting practices and safekeeping procedures employed by the custodian, the custodian's choice of sub-custodians and other intermediaries, the contractual allocation of the risk of loss, and the law governing the custody relationship. The accounting practices and safekeeping procedures employed by the custodian and sub-custodians may be the most important factors in determining the non-resident's risk of loss. Separation (segregation) of the non-resident's assets from the assets of the custodian and other customers is often the key to protecting the investor's interests.

Shortfalls in custodial holdings may develop for a number of reasons, including the failure of trades to settle as anticipated, poor accounting controls, or intentional fraud. The shortfalls may be temporary or long-standing. Allocation of the risk of loss from a shortfall will vary depending on the circumstances under which the shortfall arose. If the custodian is solvent, the risk of loss from direct acts of the custodian may be small. If, however, the custodian is insolvent, or the shortfall arises from fraud or insolvency on the part of a sub-custodian or CSD, the investor's risk of loss may be severe. In a cross-border context, the involvement of multiple legal jurisdictions and multiple settlement intermediaries increases the importance of custody risks and greatly complicates their analysis.

Another issue that increases in importance when intermediaries other than a CSD are involved is the settlement of so-called back-to-back trades and the opportunity costs and liquidity risks that arise if such trades cannot be settled efficiently. A back-to-back trade is a pair of transactions that requires a counterparty to receive and redeliver the same securities on the same day. As discussed in Section 3, securities dealers frequently need to settle back-to-back trades. However, the settlement of back-to-back trades by dealers (resident or non-resident) that are not direct participants in the local CSD poses difficulties in some settlement systems. In those systems dealers must often pre-position securities or borrow securities to meet delivery obligations. These requirements can add significantly to intermediation costs and, therefore, may significantly reduce secondary market liquidity. Securities that cannot be delivered out cannot be used to obtain secured financing. As a result, dealers incur higher financing costs and may also be exposed to greater liquidity risks, because unsecured financing tends to be less reliable, especially when financial markets are under stress. While the DVP Report highlighted the costs and risks to participants of holding cash balances and the lending of funds by CSDs to mitigate such costs and risks, it largely ignored the opportunity costs and risks that sometimes arise on the securities side when settlement of back-to-back trades is not possible and also largely ignored the lending of securities by CSDs (or, more often, by local agents) to mitigate those costs and risks.

A third issue that is quite important in a cross-border context but that was not addressed in the DVP Report is the risks associated with cross-system settlements, that is, settlements effected through links between securities transfer systems. Increases in cross-border trading and in the demand for back-to-back settlement of such trades have encouraged the development of such links. But cross-system settlements often involve significant inefficiencies that derive from the need for the transfer systems to exchange information on whether the two counterparties have the securities and funds (or access to credit) necessary to complete settlement. In particular, the settlement of back-to-back trades in which one or both settlements are cross-system settlements is often not possible, so that dealers are obliged to pre-position or borrow securities to complete such settlements.

Special problems can arise in cross-system settlements when one or both transfer systems are what the DVP Report termed model 3 DVP systems. Model 3 DVP systems make provisional transfers of securities that are not final until money settlement is completed later in the day (or on the following day in the case of systems that process instructions for settlement on S during the evening of S-1). For example, a model 1 DVP system may be linked to a model 3 DVP system. By definition, the transfers on the books of the model 1 system are final when processed. However, the transfers on the books of the model 3 system may remain provisional for many hours after they are processed. If a participant in the model 1 system receives securities across the link from the model 3 system, the status of that transfer is unclear. Either the model 1 system has received a provisional credit of securities in an otherwise final system or the model 3 system has made a final transfer of securities even though its overall processing remains provisional. This uncertainty about the finality of transfers between dissimilar systems can be exacerbated if the model 1 system permits securities received from the model 3 system to be redelivered during its own internal processing cycle. The model 1 system is then effectively granting final credit for a security that is only provisionally credited in the model 3 system.

As discussed in the DVP Report, many model 3 DVP systems provide for unwinds if a participant defaults on its settlement obligations. In such systems, a participant's failure to cover its money settlement obligation may lead to unwinds of transfers involving that participant, including transfers of securities from that participant to participants in other settlement systems. At a minimum, the unwinding of such transfers would adversely affect counterparties of the defaulting participant. However, depending on how losses are borne or allocated by the system that received the provisional transfers, others among its participants that were not counterparties to the other system's defaulting participant could also be adversely affected.

In general, the risks associated with cross-border settlements, and how significantly they differ from those involved in direct settlement in the local CSD, depend on the trading and financing patterns of non-resident counterparties and on the specific services provided by the intermediaries that they employ to hold their securities and settle their trades. Consequently, the remainder of this section considers separately the specific issues that arise when institutional investors and securities dealers effect settlements through one of the three most frequently utilised channels - settlement through a local agent, settlement through a global custodian and settlement through an ICSD.