The FSB’s Stablecoin Report: Who Holds the Coins, and Which Issuer Handles Redemption?

Cross-border coins have three different locations

On October 12, 2026, the Financial Stability Board (FSB) published a non-binding stablecoin monitoring report. When holders can redeem with any affiliated issuer, separately held reserves can face concentrated demands. Uncertain holder locations complicate monitoring. ^[1]

The starting point for understanding this problem is to draw three separate locations: the holder, the entity receiving the claim, and the reserves. Knowing a holder’s address and knowing which company receives that holder’s claim are different tasks. Adding the country where reserves are held changes the authorities involved and the documents needed. A single figure for worldwide balances cannot reveal these relationships.

Consider a company that entrusts custody to a service provider abroad. The company, the provider and the contractual recipient of a claim may each be in different places. Assigning the custody address visible on the ledger to the provider’s country does not establish where the company conducts economic activity. Using the same country allocation both for user distribution and for future redemption demand silently changes what the figures describe. The three locations inform different policy judgments. The holder’s location helps assess the currencies on which local people and businesses depend. The entity receiving claims is the unit for assessing responsibility when demand concentrates. The location of reserves helps identify which institutions may need to transfer funds. Only when the relationships between them are known can the circulation of an international coin be assessed for its effects on particular countries or financial institutions.

Related foundations: Reserves and redemption foundations — Foundations for reading reserve quality and redemption rights.

An address label is not a person’s address

Ethereum’s technical documentation describes externally owned accounts as controlled by whoever holds the private keys, while contract accounts are controlled by code. A wallet is an interface for interacting with an account, rather than the account itself. ^[4]

The questions this raises for country monitoring are practical. Is the operator also the person receiving the economic benefit? Does an address represent one company, or the many customers whose assets that company holds? Whose address should be assigned to a balance managed by code? Adding a country name without evidence connecting the technical identifier to a person or company does not strengthen the meaning of the aggregate.

Knowing the name of a custody provider need not reveal its customers’ composition. In an example involving pooled custody, companies from several countries could sit behind the balance at one address. If rights between customers are transferred solely on the provider’s books, the composition of beneficial holders can change without a movement in the publicly visible balance. This illustrates a custody arrangement; it does not describe the actual customer composition of a named provider.

Conversely, when one person uses several addresses, an increase in addresses cannot be read as an increase in users. A company might change its custody arrangements and split a balance across several places without increasing its holdings. Country aggregates therefore need a defined counting unit. People, legal entities, service accounts and technical addresses cannot be exchanged for one another whenever convenient.

Related foundations: Reading addresses and transaction records — Understand the transaction units visible on the ledger.

What research maps of transactions actually estimate

Bank for International Settlements (BIS) Working Paper 1265, published in May 2025, describes allocating cryptoasset transfers between entities such as exchanges to countries using the geographic distribution of app usage or website traffic. The researchers describe country attribution as an approximation. It is not a list of holders in October 2026. ^[5]

In assessing such estimates, consider the relationship between geographic proxies and the economic question. When frequency information is used to allocate amounts, a few large transfers could change the result. Methods emphasising frequency and methods verifying individual transaction amounts answer different questions well. The choice of proxy determines the range of uses for the estimate.

Consider retail and corporate users of the same service. An estimate allocating more transactions to a region with more browsing may help identify places where retail activity is strong. When a company makes a large payment with little browsing, browsing shares alone cannot establish the payment’s location. An estimate studying group tendencies cannot directly determine the amount of claims for which a particular legal entity is responsible. Whether an estimate is useful depends on the question. Research into the international spread of service use and an authority’s assessment of one issuer’s reserves require different levels of precision and different units of verification. There is no need to dismiss the former as having no use for the latter. But the judgment must keep clear whether the estimate is evidence of a broad trend or evidence supporting individual claims.

When a country aggregate changes, have funds moved?

Changes in classification are particularly easy to overlook when reading geographic allocations. Assigning a country to an address previously classified as unknown raises the country’s reported balance. An increase caused by coins being sent and one caused by new information about the same coins have different implications for financial markets. Treating the latter as an inflow makes better statistics look like a change in real economic activity.

To investigate this, SG Group proposes comparing the same period under two classification methods. Alongside the current aggregate produced by the new method, examine historical records reclassified under that method. Holding the input data fixed permits examination of methodological effects extending into the past. If inputs are also updated, show that difference separately. For differences remaining in the current period, investigate transactions, custody arrangements and user information. This is a procedure for investigating causes, rather than an empirical finding.

Keeping a history also matters when a service provider’s legal name changes. Counting the old and new names separately could count the same activity twice. Conversely, combining a corporate group under one name too aggressively can hide supervision and claims responsibility at individual entities. Both an identifier preserving continuity of activity and identifiers separating legal responsibility are needed. Standardising spelling alone cannot perform both tasks.

Comparable histories are useful when problems arise. They permit later investigation of whether a sharp change in country figures reflects users actually moving or a change in the provider’s classification. Deleting the old method and retaining only the latest figures obscures the basis of earlier judgments. Documents shared by businesses and authorities should therefore include the population covered and the classification version, alongside the figures.

Can an estimate establish contractual eligibility?

Circle allocates USDC redemptions to regional issuers and links its French issuer’s minimum reserves to holdings in the European Economic Area (EEA). ^[2]

The question raised by this example is how much error in country estimates is tolerable when location information also affects contractual operations. For a map of economic activity, stable broad tendencies may provide some value. For determining a claim’s recipient or allocating reserves, evidence is needed about who meets which conditions. Information estimating user activity and information establishing compliance with contractual conditions serve different evidential purposes.

When comparing location reports, investigate whether customer information changed during the period. A current balance alongside pre-relocation customer documents does not produce a comparison aligned in time. Rather than simply attaching the same date to both, explain what was checked on that date. If accounting balances and address evidence were updated on different days, retaining that distinction helps identify causes during later reconciliation.

Businesses should consequently avoid assuming that a service’s name alone explains the applicable terms. The contracting entity, definitions of location and form of custody need to be read together with the actual holder. When the operator’s nationality, the location of the device used to connect and the company’s place of incorporation differ, the contract must establish which of them matters for eligibility. This reconciliation remains necessary when using a globally shared brand.

Related foundations: How network fees work — Technical transfer costs, separate from geographic classification.

Separate location evidence according to its purpose

Monitoring information is better organised by what it establishes than placed into one “country” field in the order it is found. A customer’s declared address, the name on a contract, the place of connection and a custody provider’s registered office describe different relationships. Agreement between them improves understanding, but disagreement does not automatically establish wrongdoing. Different locations can perform different roles within an internationally active business.

SG Group’s approach is to list, for each document, the subject described, the relevant time and the judgment it can support. A custody provider’s records might describe accounts under its control without establishing every beneficiary’s address. User information can help assess holders’ locations without establishing where reserves are held. When adding evidence, also check whether the same information is merely being repeated under another name.

If documents disagree, averaging them into one country label makes little sense. First examine whether the disagreement concerns different subjects or different times. An address before and after a customer’s relocation might be explained by a history. A provider’s address and a customer’s address belong in different fields in the first place. This separates the problems without concealing missing information behind a forced estimate.

The language used for geographic categories must also match the documents. “Resident,” “located” and “established” may require different supporting records depending on whether the subject is a person or a company. Where a business sells cannot automatically replace where it is incorporated. The explanation accompanying a country aggregate is not a long caveat for preserving ambiguity; it defines the judgments for which a reader can use that figure.

Reducing unknowns and reducing misclassification are different jobs

Improving geographic monitoring involves reducing the population with unknown locations and reducing errors among those already assigned to countries. Progress in one does not necessarily improve overall reliability in the same way. The unknown category could shrink while incorrect country allocations increase. Conversely, removing incorrect allocations could enlarge the unknown category. Describing that solely as deterioration in the data would miss what changed.

It is therefore difficult to place broader estimation and stricter evidence requirements on one simple ranking. Broad estimates can provide starting points for investigating previously overlooked activity. Strict verification helps determine individual responsibility. An initial monitoring stage and the stage of handling a specific claim need not use identical evidential standards. Defining the uses of evidence by purpose preserves estimation’s role while preventing it from being mistaken for confirmed information.

Data containing only holders with known locations can also be biased. It may overrepresent businesses with complete declarations, customers using particular services or regions that are easier to verify. Their distribution need not represent all users. Even if the share assigned to countries rises, country proportions can still change if the remaining unknowns are concentrated in particular regions or custody arrangements. Separating the reasons for missing information is more useful than uniformly extrapolating confirmed cases to the world. In practice, we propose retaining reasons such as missing declarations, expired documents, conflicting names and missing breakdowns in delegated custody. Different reasons distinguish problems additional customer documents might resolve from those requiring a provider’s books. Rather than choosing a country first, identify what further evidence would change the judgment. This separates making an aggregate more detailed from making monitoring more accurate.

SG Group View: Connect the holdings map to each issuer’s burden

We consider the most useful geographic monitoring to show whose work a classification supports. Elaborating a country map without defining its purpose does not establish that information needed to determine responsibility has improved. Adding the department using an aggregate and the judgment involved makes classification improvements easier to connect to practical work. Proposals for more geographic information should be assessed by the decisions they ultimately support.

When several entities’ documents show the same custody balance, check whether their reporting populations overlap. Appearance in separate reports does not establish separate assets. Keeping holding identifiers separate from reporting entities helps match records reporting the same holding for different purposes. Retaining original rows and showing their overlap also permits reproduction when reporting scope changes later.

When assessing how current information is, examine update frequency separately from the population updated. Fresh daily figures based on unchanged, unverified customer information do not refresh the underlying location evidence. Conversely, less frequent disclosures reconciling changes in customers or custody could be valuable enough to change a judgment. This provides a comparison that avoids substituting speed for classification accuracy.

Issuers may benefit from preserving a coin’s usability while dividing operations across regions. Supporting that benefit can require expenditure on updating holder information, reconciling classifications and coordinating entities. Users and custody providers may also incur the burden of explaining the same information repeatedly. Assessing efficiency therefore requires considering who performs this verification work, alongside the apparent simplicity of a transfer.

Trace currency demand and banking effects through location

A holdings map can help assess where use of particular currencies is increasing. If a business moves part of its domestic-currency funds into coins linked to a foreign currency, its treasury management and demand for conversion could change. Seeing coins at an overseas custodian alone, however, does not establish that bank deposits left the company’s country. Separate evidence is required to connect the purchase to its funding source.

Redemption’s effects also depend on where payment is received and what happens next. A business retaining the proceeds as deposits in the same currency might not need to sell that currency immediately. Converting them into its domestic currency could create conversion demand. A record of coins redeemed for money does not choose between these outcomes. Beneficiaries’ locations need not match the locations of receiving accounts, so country balances cannot be directly converted into figures for foreign-exchange trading.

Geographic aggregates supplied by customers to banks also need methodological histories. Comparing revised figures with earlier transaction records could make a country’s relationship appear to have suddenly grown. Judgments based on that comparison risk error without an explanation of whether customer activity or population coverage changed. Companies supplying versions and revision details can help reduce banks’ reconciliation workload.

Further conditions are needed for this transmission to spread across markets. Outcomes depend on the size of claims, banks’ funding responses, reserve composition and overlap with other funding demand. The date of the FSB report is not evidence of the onset of an international banking crisis. Its market relevance lies in how information linking holder distribution to receiving issuers can help locate potential burdens.

Related foundations: Currency demand and exchange rates — Background when connecting a holdings map to prices.

Supervisory borders and corporate borders

European Systemic Risk Board (ESRB) Recommendation 2025/9 was adopted on September 25, 2025, and published in the Official Journal on November 21 that year. It addresses cross-border reserves, supervisory cooperation and data in multi-issuer arrangements involving third countries. The recommendation is not a new prohibition taking effect on October 12, 2026. ^[3]

Supervisory problems arise where corporate-group boundaries and authorities’ jurisdiction do not coincide. Even if information is combined in a group document, each authority needs defined subjects and access rights to verify individual entities. In a proposal combining international monitoring into one table, distinguish blanks caused by absent information from those caused by information held elsewhere. That identifies where further reporting is needed and where the sharing arrangements need to change.

More shared information does not invariably improve judgment. Collecting large volumes of records using different units can increase comparisons mixing users, custodians and issuers. Where institutions report the same balance within different scopes, the relationships must be reconciled before aggregation. Giving every participant detailed personal information and confirming the aggregates necessary for a decision are also different choices. When considering information sharing for a defined purpose, SG Group places weight on who can verify the basis of classification. One possible design has the entity holding identification records perform the classification while another checks the method and population covered. It separates information published openly, shared among authorities and used for individual claims. This is a design question intended to avoid both excessive collection and unsupported aggregation, rather than a conclusion about the legality of a particular system.

The case for operating without complete geographic information

There is a reasonable argument that an arrangement can operate without knowing every holder’s address, verifying eligibility when a claim arrives instead. Gathering information needed for specific claims could cost less than continually tracking all circulation. It might also be easier for users than resubmitting the same documents whenever coins move. Better monitoring need not be equated with an obligation to know every holder’s address continuously.

Assessing this argument requires separating ordinary operations from a period in which many claims arrive close together. If thorough reconciliation is possible at the claims stage, coarse prior estimates might not disrupt processing. If providers’ books or customer records are unavailable when needed, information estimating receiving workloads in advance becomes valuable. The answer concerns where reconciliation work increases, rather than collecting addresses for its own sake.

Another argument is that coarse distributions remain useful if they identify where to investigate. Studying groups with large activity changes first can be more efficient than examining every user to the same precision. Assessing this approach requires checking whether important errors remain among subjects the initial estimate did not select. If investigation concentrates solely on easily classified locations, the same bias can persist into subsequent judgments.

Evidence that would weaken SG Group’s assessment is identifiable. If eligibility for receiving issuers is consistently established, holdings with unknown locations do not obstruct claims or reserve allocation, and responses remain effective under concentrated demand, there is less reason to treat missing geographic information as a central weakness. A statement that total reserves are sufficient, however, does not establish that classification problems have been resolved. Falsification requires evidence directly testing the relationship at issue.

Current location and pre-stress history require different records

During stress, Circle SAS may temporarily cap authorised cryptoasset service providers’ (CASPs’) redemptions by balances last mandatorily reported to Circle SAS. For other EEA holders, it may temporarily restrict redemption to USDC clearly identifiable as originating from pre-stress EEA holdings, deferring the remainder until resolution. ^[2]

In assessing an arrangement using a reference date, documents describing current holdings cannot alone explain past holding conditions. Even an accurate map of the present requires separate investigation of the relationships existing at the specified time. Updating location classifications and retaining histories explaining earlier rights are tasks worth performing together. Overwriting historical records with current information removes the ability to provide that explanation.

When new information corrects an earlier classification, distinguish a record correction from a real change in the holding relationship. Correcting erroneous pre-relocation documents and updating records after a company relocates are different operations. Preserving the relevant reference time calls for separate records of the original entry, the reason for correction and the date of the actual change. Retaining history does not mean continuing to treat an old address as the current one.

When businesses explain their holdings, they can check which accounting cut-off corresponds to a reported balance and where changes in custody are recorded. If the custodian’s aggregate and the company’s books use different cut-offs, discrepancies need an explanation. Beyond aligning current balances, reproducing the population previously reported supports verification of conditions tied to reference dates. The cost of preserving records is also part of the cost of explaining future claims.

Turn agreement between aggregates into independent verification

When several aggregates agree, check whether separate information produced that agreement. If two analytics providers use the same source, their agreement does not help uncover errors in that source. Where country estimates using different methods show similar patterns, investigate the scope within which they agree. Agreement in the total does not establish agreement for a narrow population relevant to an individual entity’s burden.

Selecting cases for verification also requires attention to bias. Choosing only large, easily checked providers can verify their classifications without assessing custody arrangements with many unknown locations. Separate subjects with complete user information from delegated custody with missing breakdowns, then identify which groups retain disagreements. This gives a reason to direct resources towards finding where errors persist, alongside increasing the number of classified cases.

More specific disclosures by authorities or issuers would change this assessment. Population coverage by custody arrangement, comparisons before and after classification changes, and methods reproducing records at specified dates would make the value of additional information easier to judge. Conversely, an overall agreement rate without a description of the cases checked cannot establish whether precision is sufficient for entity-level reserve allocation. The ability to explain disagreements matters more than an aggregate’s presentation.

As stablecoins are used internationally, both technology recording transfers and documentation explaining responsibility are needed. Examining the relationship between them helps explain the significance for markets. Beyond shading a map, information that traces which evidence connects whose holdings to which issuer’s response can support judgments about international treasury management.