When to use each method

A payment from a bank in India to a bank in Singapore, where the two banks have a direct correspondent relationship and the payment is in US dollars, will usually travel serially, because there is no need to split the instruction from the funds. A payment within a banking group, where the sending and receiving banks are part of the same group and share systems, will also travel serially, because the group can move the payment internally without intermediaries. A low value payment, where the cost of managing a separate cover leg would be disproportionate, is also a candidate for serial, because simplicity matters more than speed when the amount is small. The cover method is the right choice when the sending and receiving banks have no direct relationship, when the funds need to travel through a different route from the instruction, and when the beneficiary bank needs the full information quickly. A payment from a small bank in Africa to a small bank in Asia, where neither bank has a direct relationship with the other, will usually travel by cover, with the instruction going directly and the funds travelling through a large correspondent bank that both banks use. A payment in US dollars to a country where the dollar clearing happens in New York will often travel by cover, because the funds must pass through the US correspondent while the instruction can go directly to the beneficiary bank. Currency is often the deciding factor. A payment in US dollars almost always involves a US correspondent, because dollar clearing happens in the United States through Fedwire and CHIPS. If the sending bank and the receiving bank both hold dollar accounts with the same US correspondent, the funds can move through that correspondent while the instruction goes directly, which is cover. A payment in euros within SEPA does not need cover at all, because SEPA is a single clearing system and the funds and the instruction travel together through it. A payment in a minor currency, where few banks hold nostro accounts, will often travel by cover through a major currency, usually dollars, with a conversion at one end. Amount also matters. A high value payment, say USD 1,000,000, justifies the operational cost of a separate cover leg, because the beneficiary bank needs to know it is coming and to prepare the credit, and the sending bank needs the transparency of a direct instruction. A low value payment, say USD 100, does not justify that cost, and serial is simpler and cheaper. Some banks set a threshold, for example USD 10,000, below which payments travel serially and above which they travel by cover, and the threshold is agreed in the correspondent banking agreement. The analyst should know the threshold for each corridor and should be able to explain it to operations. The analyst should not assume that a payment will always use the same method. The same two banks might use serial for one currency and cover for another, or serial for a low value payment and cover for a high value payment, depending on their agreements and on the systems available. The payment system must therefore be able to handle both methods, to generate the right SWIFT messages for each, and to reconcile the cover funds with the instruction when cover is used. The customer does not need to know which method was used, but the operations team does, because the exception handling is different. A serial payment that is stuck is traced through the chain, while a cover payment that is stuck is traced by matching the instruction to the funds leg and finding which leg is delayed. A practical decision framework for the analyst: first, is there a direct relationship between the sending and receiving banks, and if so, serial. Second, does the currency require clearing through a specific centre, such as New York for dollars, and if so, cover. Third, does the beneficiary bank need the instruction before the funds arrive, for example for a high value payment or a time critical payment, and if so, cover. Fourth, is the amount below the threshold where the cost of a separate cover leg is disproportionate, and if so, serial. Fifth, does the correspondent agreement specify a method for this corridor, and if so, follow it. That framework covers most decisions, and the exceptions should be documented, not left to habit. One more scenario worth knowing: payments inside a banking group. When the sending bank and the receiving bank are part of the same group, the payment can be moved internally on the group's own books, without using SWIFT at all for the funds leg, and the instruction can be sent as an internal message. This is sometimes called an in house or on us payment, and it is the fastest and cheapest option. The group still needs to record the payment, to screen it, and to report it, but it does not need a correspondent or a cover payment. The analyst should identify group payments in the routing table and ensure they are not sent out to SWIFT unnecessarily, because that would add cost and delay for no benefit. Worked example with real numbers: a bank in London must pay USD 500,000 to a beneficiary in Sydney, and it has two correspondents in the United States, one in New York that settles in two hours and charges USD 25, and one in Chicago that settles in six hours and charges USD 12. The payment is urgent, because the beneficiary must pay a supplier the same day, so the London bank chooses New York, pays USD 25, and the beneficiary is credited in two hours. The next day the same bank must pay USD 8,000 to a different beneficiary in Sydney, which is not urgent, so it chooses Chicago, pays USD 12, and the beneficiary is credited in six hours. The choice must be documented, because a choice made without criteria is a choice that cannot be defended. Failure and recovery: the multiple-correspondent case fails when the bank chooses the wrong correspondent, for example the cheap one for an urgent payment, so the beneficiary is credited late, and the supplier charges a late fee of USD 200, which is more than the USD 13 saved. The recovery is to pay the late fee, to apologise to the beneficiary, and to correct the selection criteria, and non-urgent payments use the cheap one. A bank that does not define urgent, for example by amount or by customer instruction, will choose wrong, and the BA must require the urgency criteria to be defined, with a default that an instruction marked urgent uses the fast correspondent, even if it costs more, because the late fee costs more than the charge saved. For a business analyst, product owner or subject matter expert, the when-to-use-each requirement must name the multiple-correspondent case, the selection criteria, urgency, cost, cut-off and amount, and the default. The ISO 20022 pacs.008 carries the correspondent in its agent fields, and the requirement must map the selection to those fields, so that a payment that should go through New York does not go through Chicago by mistake. A selection that is not documented is a selection that will be made wrong, and the test must include an urgent and a non-urgent payment on the same corridor, with different correspondents chosen for each.