Bilateral Netting in Practice

Bilateral netting is the simplest form of netting, and it involves just two parties. Two banks that exchange payments with each other regularly agree to offset what they owe each other and settle only the net difference at agreed intervals, often once a day. For example, if Bank A owes Bank B fifty million across the day and Bank B owes Bank A forty six million, the bilateral net position is four million owed by Bank A to Bank B. One payment of four million settles the entire day of activity between the two banks, instead of two separate gross payments totalling ninety six million. Bilateral netting is common in correspondent banking, where two banks hold accounts with each other and exchange payments directly without a central clearing house. It is also used in foreign exchange trading, where two counterparties may have several trades in the same currency pair and agree to net them. The arrangement depends on a clear legal agreement that defines how the net position is calculated, when settlement happens, what happens if one party fails to pay, and how disputes are resolved. Without that legal foundation, netting can be challenged in insolvency, which would defeat its purpose. The main limitation of bilateral netting is that it only works between two parties at a time. A bank with twenty correspondent relationships would need twenty separate bilateral netting arrangements, each with its own settlement and its own liquidity requirement. The total liquidity needed across all twenty relationships is still much larger than it would be under a multilateral arrangement. Bilateral netting is therefore useful and simple, but it does not deliver the full efficiency that a larger network can achieve. That gap is what multilateral netting is designed to fill. Extend the example and watch what can go wrong. Bank A owes Bank B EUR 50 million across the day and Bank B owes Bank A EUR 46 million, so the net is EUR 4 million from A to B. Now suppose a further EUR 8 million payment from A to B arrives at the netting centre two minutes after the cut off. It misses the cycle and is not netted. Bank A must now settle EUR 4 million net plus EUR 8 million gross the next day, which is EUR 12 million of liquidity it had not planned for, and Bank B receives the EUR 8 million a day later than its customer expected. A second failure is worse. Suppose Bank B cannot pay the EUR 4 million net amount on settlement day. If the two banks have a legally enforceable netting agreement, Bank A can unwind the netting and claim the gross amounts under the agreement. If they do not, and Bank B goes into insolvency, the administrator may demand the full EUR 50 million that Bank A owes while Bank A's EUR 46 million claim ranks as an unsecured creditor claim that may pay only a fraction. The netting saved EUR 92 million of liquidity, but only because the legal agreement made it enforceable. An analyst turns this into three artefacts. First, a checklist of what the netting agreement must cover. How the net position is calculated, the cut off time, the settlement deadline, what happens on a failure to pay, and how disputes are resolved. Second, an operations runbook entry for the cut off, stating what happens to payments that arrive late and who tells the customer. Third, a test case for each failure. One payment arriving after cut off, one counterparty failing to pay the net amount, and one dispute over the net figure. The test for the net figure should assert that the system recalculates the net from the gross file and raises an alert when its figure differs from the counterparty's figure by even one cent, because a small difference in the net is a sign of a large difference somewhere in the gross.