Leaving this course?

Lesson 1 of 9

An identifier that is not an account

A virtual IBAN looks exactly like a bank account number and is not one. Getting that distinction into your working vocabulary is most of the difficulty in this subject.

The European Union's new anti-money-laundering Regulation now defines it, in Regulation (EU) 2024/1624, as an identifier causing payments to be redirected to a payment account identified by an IBAN different from that identifier.

Read it twice. The vIBAN is not an account. It is a pointer. Money sent to it does not rest there, because there is no there; it is redirected to a real payment account, called the master account, which has its own, different IBAN.

What the EBA found when it looked. In 2023 and 2024 the European Banking Authority ran a fact-finding exercise on how payment service providers issue and use vIBANs, and published its report in May 2024. It recorded the common characteristics it observed: that vIBANs have the same functionality and format as standard IBANs, which makes them indistinguishable by third parties from standard IBANs, and that they are linked to a master account with its own different IBAN.

Why anybody wants them. The uses are ordinary and valuable. Give every customer of a business its own identifier and incoming payments reconcile themselves, because the identifier tells you who paid without anyone reading a reference field. A marketplace can give each seller an identifier. A payroll bureau can give each employer one. A platform can issue one per client without opening a bank account per client, which is slow and expensive. Clients get something that behaves like an account, immediately.

And why the compliance problem is structural rather than accidental. The entire value of a vIBAN is that it behaves like an account to everyone outside. That is also exactly what defeats the controls: a payment arriving at an identifier that is not an account, belonging to a party who may not be the account holder, at an institution which may not be the one holding the money, in a country that may not be where the money is.

One caution before the detail. Virtual accounts are not a European invention and not confined to the Union. The EBA report is used here because it is the most careful public examination of them, and the Regulation because it is the first instrument to define the term. The pattern is global; the citations are European.

Lesson 2 of 9

Why nobody downstream can tell

The single most consequential finding in the EBA's report is that vIBANs are indistinguishable by third parties from ordinary IBANs. They have the same format and the same functionality. Nothing in a payment message marks one as virtual.

So consider what a counterparty institution sees. A payment is instructed to an IBAN. The institution validates the format, screens the name and the identifier, decides whether the payment is domestic or cross-border, applies its own risk rules, and releases it. Every one of those steps is performed on the assumption that the identifier denotes an account at the institution whose code it carries, held by the party named. With a vIBAN, each of those assumptions may be false, and nothing tells the institution which.

The country code problem. A vIBAN carries a country code, and it need not be the country where the master account sits. The EBA found that national competent authorities took different views on whether providing a vIBAN with another Member State's country code requires establishing a branch in that Member State. So the position differs by supervisor.

For a compliance officer the implication is sharper than the legal argument. A payment may look domestic to an institution in a country where neither the provider nor the money is, which quietly changes how the payment is screened, how it is priced, what reporting it falls into, and which supervisor believes it is their business.

The one-to-one question. The EBA also found divergent views on whether an IBAN, as defined in the Single Euro Payments Area Regulation, must always be matched one to one to a payment account. If one identifier need not mean one account, then a great deal of what payment systems assume about identifiers is a convention rather than a rule, and conventions are not uniformly held.

The practical consequences, in order of how much trouble they cause.

Screening against an identifier is weaker than it appears, because the identifier may be shared, reassigned or pointed somewhere new.

The domestic-or-cross-border judgement can be wrong in both directions.

A transaction's apparent counterparty institution may have no relationship with the money at all.

And none of this is visible to the institution making the decisions, which is why the EBA treats it as a matter for the integrity of the financial system rather than a technicality between providers.

Lesson 3 of 9

The gap that had no definition

Until recently there was no legal definition of a virtual IBAN anywhere in European law. The EBA's report says so plainly: it recognises the current absence of a definition and therefore sets out the common characteristics it observed instead.

Why that absence matters more than it sounds. A supervised activity needs a name before anybody can say which rules apply to it. Without a definition, each national authority read the existing instruments for itself, and the EBA found that they reached different answers on exactly the questions that decide how a vIBAN must be controlled: whether another Member State's country code implies a branch, and whether an identifier must match an account one to one.

The report calls the result an unlevel playing field and regulatory arbitrage. Those are the right words. Where two supervisors read the same Directive differently, a provider can choose the reading it prefers by choosing where to establish, and the service it offers across the border is the same service either way.

The second kind of blindness. The EBA also found a lack of visibility for national competent authorities of the scale of vIBAN offerings in their jurisdiction, which prevented those authorities from assessing the adequacy of the arrangements. This is worth separating from the legal uncertainty, because it is a different problem with a different cure: not an interpretive dispute, but nobody knowing how big the thing is.

A supervisor who cannot count a population cannot tell whether a control is working. Neither can a compliance function. If you do not know how many vIBANs your institution has issued, to how many end users, in which country codes, pointing at which master accounts, then any statement you make about controlling them is a statement about something you have not measured.

What has changed. Regulation (EU) 2024/1624 now contains the definition quoted in the first lesson, and obligations attached to it, which lesson seven sets out. It applies from 10 July 2027, with 10 July 2029 for a few categories of business. So the definition exists and the duties are coming, and until those dates the position remains the one the EBA described.

The transferable lesson. When a product's defining feature is that it resembles something already regulated, the absence of a definition is not a gap in the law so much as an invitation. The useful question for any novel payment product is simple: what is it pretending to be, and which control relies on that resemblance being true?

Lesson 4 of 9

The identity gap, precisely

The money-laundering risk the EBA identifies is stated in one sentence worth reading slowly: the end users of vIBANs may be unknown to payment service providers, including counterpart PSPs, which means that the information used to monitor transactions may not be reliable.

Notice that the problem is reliability, not absence. A monitoring system with missing data knows it is missing. A monitoring system fed an identifier that it believes denotes a party is being told something untrue, and it has no way to know. That is worse, because every downstream inference is confidently wrong: the counterparty profile, the corridor, the relationship history, the risk score.

How the gap opens. In the straightforward arrangement, an institution issues vIBANs to its own customer and knows exactly who is behind each one. In the arrangements that cause the trouble, there are at least three parties:

  • the institution servicing the master account, which holds the money;
  • the institution or firm offering the vIBANs to end users, which has the customer relationship; and
  • the end users themselves, who may number in the thousands.

Where those are different parties, the institution holding the money may have no contractual relationship with the people whose payments pass through it. It knows its client, the provider. It does not know its client's clients, and in some arrangements it cannot.

Why ordinary due diligence does not close it. The institution servicing the master account has done proper due diligence on its customer, and that customer is a regulated payment firm with a licence. Everything about that relationship is satisfactory. The gap is not a failure of diligence on the party in front of it; it is that the parties generating the activity are one step further away, and nothing in the payment message distinguishes them.

And why the counterpart institution is in the worst position of all. It is not party to any of these arrangements. It receives a payment to an identifier it cannot tell is virtual, from or to a party it cannot verify is the account holder, and it must make a decision now. The EBA's phrase covers it exactly: the end users may be unknown to PSPs including counterpart PSPs.

The honest summary. This is a structural information failure in a chain of regulated firms, each of which has done its own part correctly. It is the same shape as the custody-chain problem in this catalogue's companion course, arriving in payments.

Lesson 5 of 9

The six indicators of higher risk

The EBA's report ends with an annex of money-laundering and terrorist-financing risk indicators for vIBANs, to be read alongside its Guidelines on ML/TF Risk Factors. It says the list is not exhaustive and that there is no expectation that every factor is considered in every case. Risk exposure is higher where these are present.

No contract with the end user. The absence of a contractual relationship between the PSP servicing the master account and issuing the vIBANs and the end users, which means the identity or location of the end user may not always be known to the institution holding the money. This is the root indicator; most of the others follow from it.

No transparency of end users' transactions. The institution sees flows through the master account without being able to attribute them to identified parties.

No limit on how many vIBANs one end user may hold. Where an end user can hold an unbounded number of identifiers, aggregation by identifier stops meaning anything, and structuring across identifiers costs nothing.

A high-risk or less-regulated country at either end. Where the master account holder, or the end user if different, is based in a high-risk non-EU country, or in a country whose anti-money-laundering rules are less stringent than the Directive's. Note that it is either end, not only the end user.

Documents that name the wrong party. Issuing documents that associate the vIBAN with names of third parties other than the verified account holder of the master account, or any feature that causes confusion about the identity of the account holder. This is the indicator most likely to be present without anyone intending harm: a statement, a confirmation or a payment-details page that shows a name which is not the account holder's.

Self-service creation without oversight. Offering customers the capacity to create, delete or deactivate vIBANs without the involvement of the PSP issuing them, with limited monitoring of their real use, for example through direct access by an application programming interface. An identifier that can be created and destroyed by the customer, at scale, programmatically, is a control problem of a different order from one issued on request.

How to use the list. Not as a score. As a set of questions to put to each arrangement you operate or receive payments from, since each indicator corresponds to a design decision somebody made and can usually unmake.

Lesson 6 of 9

What lower risk looks like

The same annex lists the indicators that may show a use case presents lower levels of risk. They are more useful than the higher-risk list, because each one is a design decision somebody can take.

A direct relationship with the end user. Where the institution servicing the master account has a direct business relationship with the end user of the vIBAN, who is identified and verified. This is the arrangement with no gap in it, and everything else on the list is a way of approximating it.

Where the servicing institution is not the one offering the vIBANs, two things reduce the risk:

  • the institution servicing the master account obtains due diligence on the end users; and
  • the institution servicing the master account and the one offering the vIBANs are based in the same Member State.

Both ends inside the Union. Where the end users and the master account are based in the EU, which removes the less-stringent-rules concern from the higher-risk list.

A regulated offeror with controls that work. Where the firm offering vIBANs to end users is an obliged entity under the Directive and has effective anti-money-laundering systems and controls in place. Note both halves: being an obliged entity is not the point on its own.

Limits on what the identifier can do. Where the institution has imposed limitations on the type of payments that can be processed through the vIBAN, the annex's own example being to top up an electronic money account. An identifier that can only do one narrow thing is a much smaller problem than one that can receive and send anything.

Only authorised agents. Where the institution servicing the master account restricts the provision of vIBANs to PSPs which are authorised agents only.

How to read the two lists together. They describe a spectrum between one arrangement and another. At one end, the institution holding the money knows the people whose payments move through it, and the identifier does one thing. At the other, an unknown number of unidentified end users in unknown countries create and destroy identifiers through an interface, and the institution holding the money sees only totals.

The question worth asking of your own product. Which of the six lower-risk features could you adopt without destroying the commercial proposition? In most arrangements the answer is at least three of them, and the reason they are absent is that nobody asked.

Lesson 7 of 9

What the Regulation now requires

The Union's answer arrived with Regulation (EU) 2024/1624, and it is worth knowing precisely, because it is drafted against exactly the gap the EBA described.

First, the definition, as in lesson one: a virtual IBAN is an identifier causing payments to be redirected to a payment account identified by an IBAN different from that identifier. Naming the thing is the precondition for regulating it.

Then the duty on the issuer. Credit institutions and financial institutions shall obtain information to identify and verify the identity of the natural or legal persons using any virtual IBAN they issue, and of the associated bank or payment account. So the institution that issues the identifier must know who uses it. That closes the gap at its source.

Then the duty that does the real work. The institution servicing the bank or payment account to which a vIBAN issued by another institution redirects payments shall ensure that it can obtain from the issuing institution the information identifying and verifying the identity of the natural person using that vIBAN, without delay and in any case within 5 working days of requesting it.

Read what that second duty actually is. It does not require the servicing institution to hold the identity data itself, which would be impractical where there are thousands of end users. It requires it to be able to get it, to a deadline. That is an obligation about contracts, interfaces and operational readiness rather than about records: if your agreement with the issuing institution does not oblige it to answer within five working days, you cannot comply, and the fix is a contractual one.

And a five-day deadline implies a process. Somebody must be able to receive the request, locate the end user behind an identifier, verify that the identification was done, and answer. An institution that has never tested that path will discover its own position the first time it matters, which will be during an investigation.

When. The Regulation applies from 10 July 2027, with 10 July 2029 for certain categories of obliged entity. Until then the EBA's description of divergent national positions stands, which means an arrangement that is acceptable to one supervisor today may be acceptable to none in 2027.

The sensible reading for anyone operating or receiving these today. The duties are known and the date is known. Building the contractual right to the identity data, and the process to exercise it, is work that can be done now and will have to be done anyway.

Lesson 8 of 9

Counting what you cannot see

One of the EBA's findings is easy to pass over and is the most portable lesson in the report: supervisors had no visibility of the scale of vIBAN offerings in their own jurisdictions, which prevented them from assessing whether the arrangements were adequate.

Why scale is a control question and not a statistic. Every judgement about proportionality depends on size. Whether an arrangement needs a dedicated control, whether a sample is representative, whether a risk is material, whether a supervisor should look: all of those need a denominator. Without one, an institution's statements about its vIBAN controls are untestable, and so are a supervisor's.

What an institution should be able to state about its own population. None of this requires new law:

  • how many vIBANs it has issued, and how many are active;
  • how many distinct end users hold them, and the largest number held by any one end user;
  • which country codes they carry, and whether the institution is established in those countries;
  • which master accounts they point to, and in which jurisdiction those accounts sit;
  • which were created through self-service interfaces rather than on request;
  • how many are associated in any document with a name other than the verified account holder;
  • and, where the institution services a master account for someone else's vIBANs, how many end users sit behind it and whether it could identify any of them within five working days.

The uncomfortable test. Pick any one of those questions and ask how long your institution would take to answer it. In most, the answer sits in a product database that no one has joined to the customer records or the payment records, which is the same finding as the custody course in this catalogue reached about settlement data.

For an institution receiving payments rather than issuing them. You cannot count someone else's vIBANs and you can do something adjacent: identify which of your incoming payment counterparties are payment institutions whose business model is issuing identifiers, and treat the apparent counterparty on those payments as what it is, a pointer, rather than as a party you have verified.

And the lesson beyond this product. Where a control's population cannot be counted, the control is an intention. The EBA's finding is that an entire market reached that state without anyone deciding it should, which is how it usually happens.

Lesson 9 of 9

Designing one you could defend

Everything above can be turned into a specification. If an institution issues vIBANs, or services a master account behind somebody else's, this is what defensible looks like.

Design choices, taken from the lower-risk indicators.

  • Limit the number of identifiers per end user, and make the limit a product rule rather than a guideline.
  • Limit the payment types an identifier can handle. The narrower the function, the smaller the problem.
  • Restrict who may offer them on your master accounts, the annex's example being authorised agents only.
  • Keep both ends in the same jurisdiction where the business allows it, and treat it as a risk decision where it does not.
  • Never issue a document or a screen that shows a name other than the verified account holder, which is the indicator most often triggered without anyone meaning to.
  • Govern self-service. If customers can create and delete identifiers through an interface, monitor the real use of them and keep the right to intervene.

Contracts and operations.

  • A contractual right to the end user's identity data from the issuing institution, with the five-working-day deadline the Regulation will require, and a tested path for exercising it.
  • A register of your own population, answering the questions in the previous lesson.
  • Attribution of flows through a master account to identified end users, as far as the arrangement allows, and an explicit record of how far that is.

What to measure. The share of vIBANs whose end user you can identify today. The time to answer an identity request end to end, measured by actually asking. The number of identifiers per end user at the top of the distribution. And the proportion of your incoming payment volume arriving from counterparties whose identifiers you cannot verify as accounts.

The honest limits. You cannot tell a virtual identifier from a real one in an incoming payment, because the EBA found they are indistinguishable by third parties and nothing has changed that. You cannot identify somebody else's end users without that institution's cooperation. And until 2027 in the Union, and longer elsewhere, the rules that apply depend on which supervisor you ask.

The sentence to carry away. A virtual IBAN is a piece of plumbing that makes reconciliation easy and identity ambiguous, and almost every control in payments rests on the assumption it quietly breaks: that the account number tells you whose money it is and where it is.

Knowledge check

Thirty questions

Answer all thirty questions, then check your answers. You need 27 out of 30 to pass and receive a certificate. If you score less, you will see which answers were right and wrong, and then go through the course again before you retake the check. Your answers, progress and times are kept only in this browser.

Sources

The official documents this course relies on. Laws and guidance change, so check the current version.

  1. Report on virtual IBANs, EBA/Rep/2024/08, May 2024 (including Annex 1, ML/TF risk indicators) · European Banking Authority
  2. Regulation (EU) 2024/1624, Article 2(26) (definition of virtual IBAN) and the identification duties on issuing and servicing institutions · EUR-Lex
  3. Directive (EU) 2015/849 (the fourth anti-money-laundering Directive), Articles 13 and 14 · EUR-Lex
  4. Directive (EU) 2015/2366 on payment services (PSD2), Article 64 · EUR-Lex