The Five Ways an FX Obligation Can Settle
The subject remains FX settlement risk, but the useful question has changed: if payment-versus-payment is the strongest protection, what happens when a trade cannot use it? This chapter compares the five settlement methods used in the 2025 BIS survey. The aim is not to rank institutions or declare one method universally best. It is to show what each method changes—the timing of payment, the amount exposed, the legal dependency or the coordination problem—and what risk still has to be managed afterward.
Start with the condition, then step down the hierarchy
The survey classifies actual two-way-payment settlement through a hierarchy of five methods: PvP, intragroup settlement, pre-settlement netting, settlement over bank accounts with timing controls, and gross bilateral settlement. The ordering is useful because it begins with a condition that eliminates FX settlement risk and then moves through methods that reduce, contain or leave that risk in place.
At the top is PvP. Its defining feature is not that a large institution stands between the parties, nor that the trade is processed electronically. It is that the final payment of one currency occurs if and only if the final payment of the other currency occurs. The two legs are linked by a shared finality condition. For the settlement leg, the principal exposure created by paying one currency and not receiving the other is removed.
The next method is intragroup settlement. A dealer bank may have transactions between desks, branches, headquarters and subsidiaries, or affiliated entities within the same banking group. The parties have strong incentives to coordinate internally, so the probability of a settlement problem may be relatively low in ordinary conditions. But the arrangement is still not the same as PvP. It relies on the group’s ability to keep its internal payments moving. In a severe stress event, a national authority might impose ring-fencing measures or a payment moratorium. Liquidity could become trapped in one entity, interrupting the sequence of internal payments and creating liquidity or credit risk that spills over to the rest of the group or beyond it.
Pre-settlement netting takes a different route. Instead of making every gross obligation move separately, it matches trades with characteristics such as the same currency and value date, then calculates a smaller net amount to be settled in each currency. If a pair of counterparties owes each other several amounts, the netting calculation can reduce the cash that must cross their accounts. That is a meaningful reduction in the amount at risk, especially when the gross obligations are large and offsetting.
But netting changes the amount, not the condition under which the remaining amount is paid. The net amount still has to be settled through another method, such as a correspondent-banking relationship, and remains subject to counterparty credit risk. The protection also depends on legal enforceability. If a netting agreement cannot be relied on when a counterparty defaults, the expected reduction in obligations may not hold in the moment when it matters most. The 2025 survey did not redistribute netted trades into another settlement bucket, because doing so would double-count them; it therefore does not reveal how each post-netting amount was ultimately settled.
Settlement over bank accounts with timing controls is a more local form of conditionality. A reporting dealer may control the timing of its outgoing payment and release its obligation only after the other party has paid with finality or posted sufficient collateral. This can be useful in relationships with other financial institutions or customers that maintain an account with the dealer. It protects the dealer’s payment decision by giving it a trigger for release.
The control can be asymmetric. A dealer may have a timing rule while the customer or other counterparty has no equivalent control over its own outgoing leg. If the parties cannot coordinate, the control may delay payment instructions rather than synchronise both legs. In stressful conditions, those delays can encourage liquidity hoarding and reduce FX trading activity. Timing controls therefore sit between full bilateral exposure and PvP: they can contain one side’s risk without creating a shared, two-currency finality condition.
At the bottom is gross bilateral settlement. The full payment amount moves through traditional correspondent-banking relationships without one of the methods above reducing or synchronising the exposure. If one currency is delivered and the other is not, the settlement-risk exposure is the full trade value. This is the baseline against which the other methods should be understood—not a moral label attached to a bank or trader.
Why the middle methods are not interchangeable
The hierarchy becomes more useful when the methods are compared on the same four questions: Does the method synchronise the two currency legs? Does it reduce the amount that must move? What legal, operational or liquidity conditions does it depend on? What still has to happen after the mitigation is applied?
These questions prevent a misleading comparison based only on whether a method sounds safer. Imagine the same pair of counterparties with several obligations in both directions. If an eligible PvP path is available, the final currency legs can be linked. If the obligations can be legally netted but the currencies or product cannot enter PvP, the gross amount may be compressed before a smaller amount moves through correspondent accounts. If the trade is internal to a banking group, internal coordination may be the practical control. If a dealer can release funds only after a customer pays or posts collateral, a timing control may reduce the dealer’s exposure. The underlying trade has not changed simply because the settlement method has; what changes is the part of the payment sequence the method controls.
A simple illustration makes the difference visible. Suppose two counterparties have several offsetting claims that would require $100 million to move in one direction and $98 million in the other. Netting could leave a $2 million amount to settle, reducing the gross amount that would be exposed. It would not, by itself, determine whether that $2 million is exchanged against the other currency at the same moment. PvP could supply that condition if the netted payment were eligible and entered the system. The two methods can therefore complement each other, but neither should be described as doing the other’s job.
The same logic applies to intragroup settlement. Internal coordination can make a payment path more reliable in normal conditions because the group can see and manage its own positions. Yet the entities still operate through legal and liquidity arrangements that may become fragmented in a crisis. A shared PvP condition protects the exchange between two currencies; intragroup control protects a set of internal relationships only as far as the group can keep those relationships funded and legally connected.
PvP is the only method in this hierarchy whose defining condition synchronises final payment of both currencies. It can also enable payment netting, which improves liquidity management, but its essential protection is the if-and-only-if rule. The method works best as a shared settlement path for a trade that the relevant participants, currencies and product can enter. Where it is available and used, it directly addresses principal risk.
Intragroup settlement reduces reliance on an external counterparty by placing the transaction inside a banking group’s internal coordination structure. That may make timing easier to manage, but it does not make internal entities one indivisible pool of liquidity in every legal or political circumstance. A branch or subsidiary can be subject to a local restriction even when the wider group remains solvent. The method’s strength is coordinated control; its failure surface is the possibility that the group’s liquidity and payment sequence become fragmented.
Netting reduces gross value. Its benefit is clearest when many obligations offset one another, because the final amounts can be much smaller than the original trades. Its failure surface is different from intragroup settlement’s: the critical questions are whether the agreement is enforceable on default and how the remaining net amount is paid. Netting can be a valuable bridge where PvP is unavailable or too costly, but it cannot be described as equivalent to PvP merely because both methods lower the amount exposed.
Timing controls reduce the chance that one participant sends funds before the other has paid. Their effectiveness depends on coordination between counterparties. A control held by only one side may protect that side while imposing delay on the relationship as a whole. The resulting trade-off is important: a bank may prefer to preserve cash until it sees final incoming funds, but widespread caution can slow the market and encourage participants to hold liquidity rather than circulate it.
Gross bilateral settlement does none of these things. It does not synchronise the legs, reduce the gross amount through the settlement method, rely on an internal group relationship or hold an outgoing payment behind a shared control. It is therefore fully exposed to settlement risk for the value delivered. That description is about the mechanics of the method, not an accusation that every bilateral trade is imprudent. A trade may be gross bilateral because the stronger method is unavailable, the legal arrangements are incomplete, the timing does not fit or the counterparties cannot connect to the relevant infrastructure. The method remains exposed even when the decision to use it is understandable.
The methods also differ in what they leave unknown. PvP gives a clear condition for final payment. Intragroup settlement depends on the group’s internal liquidity and legal environment. Netting tells us that the gross amount has been compressed, but the survey does not tell us the subsequent settlement method of the net amount. Timing controls tell us that one dealer conditions its payment, but not that both counterparties have equivalent controls. Gross bilateral settlement tells us that no mitigation method was applied to the full amount.
This is why “mitigated” should not be read as one uniform category. In the 2025 survey, intragroup settlement, pre-settlement netting and timing controls were grouped as methods that reduce—but do not eliminate—FX settlement risk. They reduce risk in different ways and leave different dependencies behind. A group may need mobile liquidity across entities. A netting arrangement may need enforceable legal treatment and a later payment channel. A timing control may need the other party to coordinate or provide collateral.
The hierarchy is therefore a guide to the kind of protection in use, not a promise that the second method is always safer than the third in every transaction. A netted payment may be much smaller than an intragroup obligation, while an intragroup relationship may be easier to coordinate in normal conditions. The relevant comparison depends on the failure being considered. If the concern is an unpaired final payment, PvP is the direct solution. If the concern is the amount that must be funded, netting may provide the larger immediate reduction. If the concern is a dealer releasing funds to a customer, a timing control may address the dealer’s particular exposure while leaving a wider coordination problem.
The choice can also be staged. A market participant may use PvP for eligible trades, netting for obligations that cannot yet enter the system, and internal or timing controls for relationships where those arrangements are practical. That is not a contradiction in the hierarchy. It is an acknowledgement that settlement protection is built from interfaces between participants, currencies, products, legal systems and payment schedules. Where PvP is impractical or too costly, robust netting can be a shorter-term mitigation, but the residual settlement dependency should remain visible.
There is also no single method that dominates every operational situation. PvP is the strongest answer to principal risk when a trade can enter the protected path. Netting may be useful for currencies, products or schedules that do not fit that path. Intragroup settlement can reflect the real organisation of a dealer’s internal funding, while timing controls can be a practical safeguard in a customer relationship. These choices do not erase the hierarchy; they explain why the hierarchy exists.
The important distinction is therefore between the best available method for a particular trade and the amount of residual risk that method leaves. The hierarchy gives us a disciplined vocabulary for making that distinction. It also prepares the quantitative question: how much of the April 2025 market used each method, and where did the gross bilateral remainder cluster? The answer will show that the remaining exposure is not spread evenly across FX. It is concentrated at the boundary between what settlement infrastructure can accept and what the market is actually trying to settle.