The default waterfall: who pays when a participant fails

The default waterfall is the ordered list of resources that a payment system uses to cover the loss when a participant fails to settle, and it is the last line of defence before the loss falls on the surviving participants. The name comes from the way the resources are consumed in sequence, like water falling from one level to the next: first the failing participant own resources, then the system default fund, then assessments on the surviving participants, and finally, in some systems, a central bank backstop. The waterfall is defined in the system rulebook, and every participant agrees to it when it joins. The PFMI requires that the waterfall be sufficient to cover the default of the largest participant, and the operator must test this regularly. Why the waterfall exists is that a settlement failure creates a real financial loss, not just a delay. If Bank A fails to pay its EUR 25 million net debit, the banks that expected to receive that money are EUR 25 million short. Someone must bear that loss, and the waterfall decides who. Without a waterfall, the loss would fall arbitrarily on whichever banks happened to be owed money, which would be unfair and would discourage participation. With a waterfall, the loss is allocated in a known order, the participants can price the risk, and the system can continue operating. The Bank of England and the European Central Bank both require their systems to have a waterfall that is transparent and tested. The parties are the defaulting participant, the system operator, the surviving participants, and in some systems the central bank. The waterfall typically has four levels. First, the defaulting participant margin or collateral, which is money it has already posted to the system. Second, the default fund, which is a pool of money contributed by all participants in advance. Third, an assessment, which is a call for additional money from the surviving participants, usually capped at a multiple of their default fund contribution. Fourth, in some systems, a central bank liquidity facility or a loss-sharing agreement. Each level is consumed in full before the next is touched, and the order is fixed in the rulebook. In practice the waterfall is applied like this. Bank A defaults on a EUR 25 million net debit. The operator first uses Bank A margin of EUR 5 million, leaving EUR 20 million. It then draws EUR 12 million from the default fund, leaving EUR 8 million. It then levies an assessment on the surviving participants: if there are four survivors and the assessment is pro rata to their default fund contributions, each pays EUR 2 million. The loss is fully covered, and the system settles. The default fund must then be replenished, usually within a defined period, so that the system is protected against the next default. The operator publishes a default report showing how the waterfall was applied, so that every participant can see what happened. A worked example shows the arithmetic. The default fund holds EUR 50 million, contributed by ten participants at EUR 5 million each. Bank A defaults on EUR 25 million. Its margin is EUR 5 million, so EUR 20 million remains. The default fund covers EUR 20 million, falling to EUR 30 million. No assessment is needed, because the fund was sufficient. But if Bank A had defaulted on EUR 60 million, the margin of EUR 5 million and the entire default fund of EUR 50 million would be consumed, leaving EUR 5 million, which would be covered by an assessment of EUR 500,000 on each of the ten surviving participants. The assessment is capped in the rulebook, typically at one or two times the default fund contribution, so no participant faces an unlimited call. What happens when the waterfall is exhausted is the extreme case that the PFMI calls a uncovered loss. If the default is so large that the margin, the default fund and the maximum assessment together cannot cover it, the system must either suspend settlement, which freezes the whole market, or the central bank must intervene. This has never happened in a major payment system, but it is the scenario that stress tests are designed for. The operator must run a stress test at least annually, simulating the default of the largest participant under stressed market conditions, and must publish the results to the participants and the regulator. If the test shows the waterfall is insufficient, the operator must increase the default fund or the assessment cap. The operational and accounting implications are that the finance team must manage the default fund contributions, the risk team must monitor the largest exposure, and the operations team must be ready to execute the waterfall at short notice. On the ledger, a default fund contribution is an asset of the participant, and a draw on the fund is a loss that must be recognised. An assessment is a liability that must be paid within the deadline in the rulebook, usually one business day. The accounting team must also track the replenishment obligation, because the default fund must be restored to its full size before the next settlement cycle. For a business analyst, the default waterfall is a rules and reporting subject. You must specify the levels of the waterfall, the size of the default fund, the contribution formula, the assessment cap, the replenishment deadline, and the reporting. You must also specify the stress test scenarios and the frequency, and the procedure for notifying participants when the waterfall is used. A common gap is failing to specify the replenishment obligation, which leaves the system unprotected after the first default. Takeaway: the waterfall turns an unpredictable loss into a known allocation, and the size of the default fund is the system risk appetite expressed in money. Further reading: the BIS Principles for Financial Market Infrastructures (PFMI) for the default waterfall and stress testing; the Bank of England Financial Market Infrastructure supervision; the ECB TARGET default fund arrangements. Last reviewed 7 November 2026.