BIS figures place $1.4tn of daily FX settlements in the exposed gross-bilateral channel

The foreign-exchange trade that matters most to a risk manager is sometimes the one that has already been executed successfully. Price has been agreed, the confirmation is in place, and the counterparties know which currencies to deliver. The final question is whether each actually receives the currency it is owed. A new analysis of the Bank for International Settlements' 2025 Triennial Survey puts that unfinished business into proportion: more than US$1.4 trillion of average daily FX obligations were settled on a gross bilateral basis without the risk-mitigation methods that protect much of the rest of the market.[1]

That is not a claim that US$1.4 trillion was lost, uninsured or left unsecured on a balance sheet for a whole day. It is a measure of the gross value of settlement instructions using a method that exposes one counterparty to the possibility of paying away its currency before receiving the other. The distinction is indispensable. Actual defaults, expected credit losses, gross settlement flows and notional trading turnover are four different concepts. The important original finding in our reconciliation is how the at-risk proportion changes when dealer intragroup flows are stripped out and when the smaller, better-hedged netting category is examined separately.[1][2]

BIS counted more than US$14 trillion of average daily gross financial settlement obligations in April 2025. Its classification assigns approximately US$5.2 trillion, or 36%, to payment-versus-payment systems; around US$7.6 trillion, or 54%, to methods that mitigate but do not eliminate settlement risk; and more than US$1.4 trillion, or 10%, to gross bilateral settlement. Those amounts are rounded survey estimates, so their sum should not be used to manufacture a more precise global denominator than the source supplies. The three categories are not an allocation of annual trading income or a census of individual retail broker client balances.[1]

The point at which a currency trade becomes a credit exposure

In a payment-versus-payment, or PvP, arrangement, the final transfer of one currency is conditional on the final transfer of the other. The point is not to make the buyer or seller more creditworthy; it is to prevent the particular asymmetry in which the first paying party irrevocably sends funds while the other can fail to perform. The BIS describes PvP as eliminating foreign-exchange settlement risk for trades that settle successfully through that mechanism. It does not promise to eliminate every liquidity, operational or market risk associated with trading.[1][2]

In gross bilateral settlement, the full amounts are exchanged through the counterparties' payment arrangements without equivalent settlement-risk protection. The currency pair may not be eligible for an available PvP service; the counterparty may lack access; or the transaction type may not fit. A firm can still have counterparty limits, collateral arrangements, strong operational controls and a high-quality bank on the other side. Those may reduce the probability or severity of loss. They do not change the fundamental sequence: settlement of one leg is not mechanically conditional on the other.[1]

The other 54% should not be casually called 'safe'. Intragroup settlement moves obligations within a banking group, reducing some external credit exposure while leaving questions about legal entities, funding and liquidity under stress. Pre-settlement netting compresses offsetting trades into smaller payments, but the remaining payments still have to be delivered. Bank-account methods with control over settlement timing help manage the sequence without necessarily locking both currencies together. Those are different risk reductions, so counting them all as the same thing would make a misleading comparison with genuine PvP.[1][3]

The Global Foreign Exchange Committee's settlement methodology presents this hierarchy explicitly. Its first preference is to eliminate settlement risk with PvP where practicable, then to reduce the value and duration of any residual risk, with gross bilateral settlement treated as the method offering least protection. The Code is a voluntary set of good-practice principles and not itself a statute imposing identical obligations in every jurisdiction. That matters when translating a global survey finding into an allegation about one broker: a higher-risk method is a reason to investigate exposures, not proof of a rule breach.[2][3]

Intragroup payments change the denominator

The survey says about 35% of average daily gross financial obligations settled within dealer groups. It also reports that the interdealer category generated roughly US$9.3 trillion of total settlement obligations, of which more than US$4.9 trillion was intragroup. After those internal movements are removed, external interdealer settlements were around US$4.3 trillion. About US$474 billion of that external amount was settled on a gross bilateral basis. Dividing 474 by 4,300 gives an indicative 11.0%, compared with roughly 5% when the much larger, internal-inclusive interdealer base is used. Both percentages can be correct because they answer different questions.[1]

A superficial review might conclude that the interdealer market's gross bilateral problem is tiny because only one in twenty dollars is assigned to it when internal movements are included. For a manager focused on unsecured transfers between legally separate external firms, that denominator is not the relevant one. The external-only view raises the at-risk share to about one in nine dollars, still below the corresponding shares reported for other financial institutions and non-financial corporate counterparties, approximately 19% and 22% respectively. These are shares of settled obligations, not shares of customers suffering loss.[1]

The contrast says something substantive about infrastructure access. Large dealer banks often maintain relationships with CLS and other settlement services, whereas an asset manager, corporation or smaller participant may use a correspondent bank and the settlement terms its providers offer. A company hedging a supplier payment and a global dealer offsetting a book can trade the same currencies but reach different settlement rails. The survey's percentages suggest that the ability to use a protected method is not distributed evenly. The data do not name individual counterparties and cannot establish which company refused protection or why.[1][2]

Netting is powerful, but the outstanding number matters

The BIS reports that pre-settlement netting reduced approximately US$2.2 trillion in average daily gross obligations to around US$337 billion in actual remaining amounts to settle. The arithmetic reduction is about US$1.863 trillion, or 84.7% of the original gross amount. This is a genuine risk-management achievement. It does not mean that 84.7% of every counterparty's potential credit loss disappeared. Netting compresses payments under agreed rules, while the remaining bilateral currency transfers still carry risks according to how they are settled.[1]

The often-neglected limitation is that the BIS did not collect a breakdown showing how the US$337 billion net amount was divided among interdealer, other financial and non-financial counterparties. A report claiming that customers themselves retained a particular dollar value of post-netting risk would therefore be adding an unsupported fact. The counterparty proportions are available for some settlement categories, but they cannot simply be superimposed on the net amount. This is a particularly important caution for brokers using wholesale FX liquidity: the apparent low number after compression is not a disclosed client exposure measure.[1]

Netting is also dependent on documentation and legal enforceability. If two counterparties have obligations in the same currency on the same value date, a properly supported agreement can reduce the number and size of transfers. It does not follow that unrelated entities inside the same corporate group can freely offset all positions without a relevant agreement, or that offsetting trades with different settlement dates can be collapsed without further arrangements. Those constraints are why the Global FX Code stresses agreed methods and consistent application rather than treating the arithmetic of offsetting tickets as a substitute for risk controls.[2][3]

Trading volume is not the same as money to be delivered

The BIS's separate 2025 turnover statistics put average daily OTC FX trading at US$9.5 trillion on a 'net-net' reporting basis. That turnover number is not a valid denominator for calculating the share of settlement obligations at risk. It uses a different methodology and counts market activity rather than the gross financial obligations and payment chains covered by the settlement survey. The same economic risk transfer can generate several transactions and settlements; two legs may fall on different value dates; some transactions create derivative cash flows rather than a simple spot currency exchange.[1][4]

The distinction helps explain why a settlement obligation total greater than the trading turnover headline is not automatically a contradiction. Different units of account, internal movements, settlement methods and reporting adjustments produce different aggregates. The correct question is what each study measures. The trading survey reports activity by sales desks under its conventions. The settlement analysis reconstructs amounts that have to be paid. Treating the two as interchangeable would generate a sensational but false percentage and could even double-count operational risk.[1][4]

The same trap arises when comparing CLS's own reported daily activity with the BIS survey. A settlement service can announce values processed or submitted under its own definitions, and those figures may be invaluable for operational understanding. But a service's activity measure does not, by itself, allow us to calculate the share of all global FX trades protected by PvP without harmonising transaction types, dates, counterparties and double-counting rules. The BIS survey provides the appropriate cross-market comparison here.[1][2]

A small failure rate and a large gross exposure can coexist

The survey separately measured unsettled amounts remaining at month-end for transactions due to settle during April 2025. It found settlement fails equal to about 0.01% of gross financial obligations settled, with approximately US$121 million of fails among reporting dealers and US$1.4 billion in the other non-reporting financial-institution category. Those are different concepts and periods from the US$1.4 trillion average daily gross bilateral settlements. The fact that the numbers happen to begin with the same digits does not establish any causal relationship.[1]

The reported fails show that the vast majority of surveyed settlements were completed, but a low realised failure rate is not the same as a low potential severity if a major counterparty becomes unable to perform. A risk manager should consider the interval between the first and second currency payments, liquidity needs, replacement costs and correlations between counterparties. A method that works reliably in calm conditions can still create concentrated exposure in stressed markets. These are risk-management implications of the BIS classification, not a measurement of future default losses.[1][3]

There is a further historical trap. The BIS cautions that its revised 2025 methodology cannot be compared directly with its 2019 and 2022 survey results on settlement risk. It offers a separately constructed comparison with an earlier 2006 survey. One cannot take an old pie chart, subtract its 'gross bilateral' slice from the 2025 number and announce that the industry's risk fell by a particular percentage unless the definitions and coverage are truly aligned. The safest reading is the current distribution and the expressly documented methodology, not a spurious multi-decade annual trend.[1]

What should be disclosed by liquidity providers and brokers

A broker buying external spot or forward liquidity should be able to identify which legal counterparty settles each trade, which currency pair and value date are involved, the status of netting agreements, and whether PvP is available. Its own retail customers need not be directly involved in each wholesale settlement transaction. Their potential exposure depends on the broker's capacity, cash and liquidity arrangements, and whether an external settlement failure affects its ability to manage or settle customer obligations. This is why it is crucial not to re-label bank-level settlement obligations as retail customer money.[1][2][3]

A meaningful operational report would separate gross obligations, amounts compressed through netting, net payments actually made, eligible but unused PvP opportunities, balances outstanding past value date and recovery times after a fail. For each exception it would record the counterparty legal entity and reason the preferred method was unavailable. These are recommended controls derived from the public measurement framework, not a claim that this reporting inspected individual brokers' back offices.[1][3]

The main new perspective is therefore less dramatic than the largest number and more actionable than a general warning about counterparty risk. On a global basis, 90% of the reported gross settlement value was subject to a method that eliminated or mitigated settlement risk, while around 10% still used the exposed gross bilateral method. But the external-dealer denominator, the 84.7% netting compression and the non-dealer participation gap show why a single market-wide percentage is inadequate for judging any one liquidity network. The next advance will come not from repeating US$1.4 trillion, but from making the legal entities and settlement method choices behind that number more transparent.[1][2][3]

Report an error or submit an update