Andres Knobel ■ New report on how to fix beneficial ownership frameworks, so they actually work

Magnifying glass above a coin

Close to 100 countries have approved beneficial ownership registration frameworks. Many of these countries will likely have to upgrade them after the recent reforms of the Financial Action Task Force recommendations 24 and 25 on beneficial ownership transparency for legal persons and trusts. And if not based on the reforms, then based on the pretty bad ratings given them by the Global Forum 2023 Annual Report:

EU countries, which are still reeling from the European Court of Justice’s ruling against public access, continue to negotiate changes to their beneficial ownership registries as part of the so-called “AML Package”. In this context, the Tax Justice Network has published a new report “Why beneficial ownership frameworks aren’t working – and what to do about it”.

Why this report?

We agree with secrecy supporters that beneficial ownership registries haven’t solved the issue of illicit financial flows. Tax evasion, money laundering and corruption keep flourishing unabated. However, this is not a reason to throw loophole-ridden beneficial ownership registries out with the proverbial bathwater. On the contrary, it just adds more urgency to get them right.

There are many things to fix. Applying verification and expanding access to all stakeholders (ie public access) so that beneficial ownership can serve multiple uses are obvious first steps. But our latest report deals more with the fundamentals: getting the law right to begin with. In this regard, complying with the weak standards of the Financial Action Task Force or the Global Forum won’t solve things either. As we’ve long warned, these standards have advised governments to build houses on sandy soil – it’s shouldn’t come as surprise then when walls begin to crumble.

Beneficial ownership vs legal ownership 

Most companies have very simple structures where the legal owner (ie shareholder) directly owns and controls the company. In this scenario the shareholder is also be the beneficial owner. This information is publicly available in most commercial registries. That’s all good. The problem starts with complex ownership structures – ie where a company or legal entity is owned by many layers of entities spread across tax havens. These structures have been used by oligarchs, criminals and tax abusers to circumvent the rule of law. When it come to these structures, relying on just commercial registries is usually not enough to solve the problem. 

Consider the following complex structure, where each colour represents a different country of incorporation and where the real beneficial owners are Mary, John, the settlor and the beneficiary of the trust sitting at the top of the structure:

Relying on only legal ownership information held in commercial registries wouldn’t be enough to identify all these beneficial owners, even if (and this is a very big “if”) information on all the real owners was  available in all relevant commercial registries. 

The commercial register of country red would reveal that Company 2 is the legal owner of Company 1 (Mary wouldn’t appear in the commercial register). The next commercial register (country blue) would identify Company 3 and John as owners. The commercial register of country green would identify the trustee as shareholder (there might be no indication that the trustee is acting as a trustee on behalf of the trust, and could give the impression that she is the real beneficial owner). Even if the trustee declared the name of the trust to the commercial register of country green (which is unlikely), it would be even more unlikely that country orange has a trust register in which to find information on the settlor and the beneficiary. After a very time-consuming search, an investigating authority will at most identify just John and the trustee.

Current beneficial ownership registries aren’t that much better. Many of them include basic legal ownership information (eg Company 2) but not the full ownership chain (Company 3 and the trust) needed to confirm the declared beneficial owner on top. As for the beneficial owner information they hold, they would likely require the registering of the trustee, and in the best case scenario the settlor and beneficiary as well. But, Mary, who merely has influence (which may be hard to detect without advanced verification mechanisms) would likely avoid registration requirements, as would John. Only a truly effective beneficial ownership register would be capable of providing an overview of the whole ownership chain as shown in the figure.

Why beneficial ownership frameworks are flawed (even if in compliance with international standards) and how to fix them

As the brief describes, following the Financial Action Task Force’s recommendations, most countries identify as beneficial owners only those individuals who hold more than 25 per cent of the capital in a legal entity, and where no one is identified, those with “control via other means”. 

 Based on the goal of saving costs for companies and for firms, financial institutions and other obliged entities that must undertake customer due diligence, countries take a “small” data approach (rather than the “right-data” approach), asking for as little information as possible and hoping that this makes it easy and cheap to collect and verify information. There’s perhaps also a naive hope here that that the catch-all phrase “control via other means” will capture all relevant individuals while asking for little.  This supposedly cost-saving choice is a false economy and comes at a great cost: authorities are unable to obtain all the information they need, instead needing to spend considerable time to identify the real beneficial owners.

Instead, the brief explains that a no-threshold approach (ie requiring the identification of any beneficial owner with at least one share or vote), coupled with more extensive details to be collected (eg information on those with a power of attorney or with exposure to a company’s economic performance) is the only way to ensure that authorities will have all the beneficial ownership information they need when the need arises. This way, authorities can readily analyse the information they hold to reveal undisclosed relationships, properly conduct investigations into crimes, and detect crimes that might otherwise go unnoticed.

Importantly, as this brief will show, implementing a no-threshold approach does not add costs: even a framework with a 25 per cent threshold presupposes that anyone holding directly or indirectly at least one share has been identified. This is the only way to aggregate all holdings and determine which of those passed the threshold. Switching to a no-threshold approach would require no additional work in identifying direct or indirect shareholders. 

Busting the claims against a no-threshold approach

The sandy soil governments were advised to build their beneficial ownership frameworks on is the “25 per cent or more” threshold. The single most effective action governments can take to strengthen their beneficial ownership frameworks is switch to a no-threshold approach, following the examples of Argentina or Ecuador. This would identify as a beneficial owner any individual who has at least one share or vote in a legal entity – and eliminate the transparency escape hatch that is making beneficial ownership registers ineffective today.

Opponents to the no-threshold approach have made some arguments pushing back against the approach. The brief and the table below responds in detail to these. 


Claim: Requiring “all” individuals to be registered (because no thresholds are applied) is by definition disproportionate.

Our response: There are plenty of cases where regulations apply to “all”: all legal owners must be identified with the commercial registry, all parties to a trust must be identified as beneficial owners. In many countries, all individuals must obtain a national ID, all taxpayers must file tax returns, all individuals wanting to drive must obtain a driving license, etc.

A measure should not be automatically considered “disproportionate” just because it applies to “all” people or companies rather than to specific ones. Collecting information about all individuals with a link to an entity (rather than only of those with more than 25 per cent of the shares) is not necessarily disproportionate.

The goal of beneficial ownership frameworks is to tackle illicit financial flows such as corruption, money laundering, the financing of terrorism, tax evasion, sanctions evasion, etc. Collecting beneficial ownership information for all owners regardless of the value of their shares does not impose a burden on an individual that is excessive in relation to the objective sought to be achieved. In fact, a no-threshold approach already applies in the case of legal owners of companies (all persons must disclose their direct shareholdings to a commercial register) and to beneficial owners of trusts (all parties to the trust must be identified as beneficial owners regardless of their interest or rights to the trust assets and income). In fact, for most companies with simple structures where the beneficial owner directly holds the shares, beneficial owners are already identified with a no threshold approach – eg in the UK more than 80 per cent of companies have simple structures.

The only ones who would be affected by a no-threshold approach are those who intentionally create complex offshore structures to remain below thresholds, as a way of avoiding transparency of their financial affairs.

In addition, most administrative processes involve an “all” approach. In many countries, all individuals must obtain a national ID, which may include providing their date of birth, address, signature and fingerprints. This does not mean that all individuals are regarded as potential criminals. This is simply information that the state needs to fulfil its obligations, including the prevention of crime, ensuring economic fairness, planning and budgeting for social services, identifying missing persons, etc.

In many jurisdictions, all taxpayers must file tax returns, not because they are all considered tax evaders, but because it is how tax authorities ensure and verify compliance. In addition, having information on all taxpayers, both honest and not, allows authorities to compare them, to find patterns or red flags to ensure a level playing field where everyone pays their fair share. All customers must provide a financial institution with information for the know-your-customer due diligence procedures, not because they are would-be money launderers but in order to apply proper checks.

A risk-based approach allows for additional measures to be taken, but for it to be effective in detecting anomalies and outliers it does require obtaining a minimum amount of information from all.


Claim: A definition without thresholds does not pass the necessity test because an individual with 1 per cent or less of the shares would not be in control of the company and would thus not be responsible for any crimes.

Our response: Thresholds are deliberately exploited by criminals and those who want to remain hidden from authorities, undermining the whole purpose of beneficial ownership transparency. A person with even a 0.01 percent interest in a listed company (eg Apple) would have no control whatsoever over the entity, but that that tiny percentage could be worth millions of dollars. Identifying this person could be relevant to money laundering, corruption, tax evasion, etc.

Criminals exploit loopholes, especially thresholds, to remain hidden from authorities. 

Beneficial ownership transparency is about identifying the real owners and controllers of legal vehicles to prevent them from engaging in illicit financial flows such as corruption, money laundering, tax evasion, etc. Thresholds are deliberately exploited by criminals and those who want to remain hidden from authorities, undermining the whole purpose of beneficial ownership transparency. Thresholds allow individuals to remain hidden, either by distributing their interests so they are slightly below the threshold, or directly by not having any ownership interest but rather holding control through a power of attorney, financial instruments, etc. As visually illustrated in a publication on beneficial ownership and investment funds , a 0.01 per cent interest in Apple would give no control whatsoever over the design of the iPhone. However, that tiny percentage would be worth more than US$200 million. Identifying the beneficial owner of that 0.01 per cent would be relevant to determine whether taxes have been paid and how the beneficial owner afforded that interest to begin with, to dispose of cases of corruption or money laundering.

A no-threshold approach is the only way to ensure all relevant individuals are covered, no matter their circumvention attempts.

Presumption of innocence

Claim: Preventive collection of data and pattern-finding violate criminal law principles. Looking for patterns to investigate specific individuals before suspicions exist violate the principle that “the suspicion leads to the investigation” rather than the other way around.

Our response: Crime prevention is just as important as its prosecution. There is no need to wait until a criminal act or wrongdoing happens in order to act. Crime prevention does not affect the presumption of innocence. Most legal frameworks put a lot of emphasis on prevention, not because they consider all individuals as future criminals or victimisers but to prevent them from becoming such. 

All drivers need to obtain a licence to prove they know how to drive, where their sight and hearing is also tested. Seatbelts are compulsory. Drinking alcohol and the use of cell phones while driving is prohibited. Cars have licence plates so they can be identified. None of these rules can be interpreted as an infringement on the presumption of innocence. At the same time, all of this information could be relevant in case of a car accident. In the same way, obtaining information on all beneficial owners related to an entity, checking that they are not related to any criminal or that there aren’t other red-flags (similar to checking a driver’s sight and hearing) cannot be considered an infringement of the presumption of their innocence, or as impugning their good faith or honesty. 

This is somewhat similar to how airport security helps to prevent acts of terrorism, except in this case by preventing the economic and human cost of financial crimes. Both airport security and beneficial ownership transparency apply to individuals who are not viewed as either criminals or terrorists. One of the arguments against public access and collection of beneficial ownership information on all individuals related to an entity is that it affects compliant and honest citizens who have done nothing wrong. A simple counterargument is that the vast majority of airplane passengers are not terrorists, but everyone is still required to go through airport security. This comes at some cost in terms of the additional time spent by passengers as well as the staff and infrastructure required – but the cost is offset by the prevention of greater, even more costly harms.

One of the likely reasons why millions of people agree to or at least accept the discomfort of airport security (which includes affecting their privacy as every item of their luggage can be checked) is the immediate relationship between terrorism and the loss of life. By contrast, beneficial ownership transparency is perhaps perceived as a less important, urgent or worthy “privacy-affecting” measure because it has a more indirect link to the violation of human rights. However, the fact that the link to beneficial ownership transparency is more indirect does not mean that it is irrelevant.

Although the link between beneficial ownership transparency and human rights violations is indirect, the consequences and effects can be much broader, when considering all the financial crimes and unfair situations caused by secrecy. In economic costs, the State of Tax Justice in 2023 estimated that countries will lose US$4.8 trillion in tax to tax havens over the next 10 years. The UN estimated in 2018 that the global cost of corruption was at least US$2.6 trillion, or 5 per cent of the global gross domestic product. All of these resources are critical to pay for fundamental human rights – health food, education, housing, a fair trial, and many others.

Although harder to quantify, financial crimes such as corruption also have a human cost. Corruption led to the deaths of 52 people in 2012 in a factory fire in Bangladesh, just as it did for almost 200 young people who were trapped at an illegally held music concert called “Cromañon” in Argentina in 2004. More than 40,000 people died in the 2023 earthquake in Turkey due to poor construction regulation and corruption, and Lebanon’s port explosion killed more than 200 people due to illegally stored chemicals.

Trust in government

Claim: Some human rights organisations and activists distrust what they believe governments will really do with the collected information, eg target vulnerable populations or political opponents.

Our response: State authorities aren’t necessarily the problem. Oligarchs, high net worth individuals and entities who are often more powerful than countries, pose a bigger risk to democracy, equality and the rule of law. A no-threshold approach is a way to protect minorities and vulnerable individuals by ensuring that powerful individuals won’t be able to escape the rule of law by creating complex ownership structures. 

Criminal law, constitutions or human rights conventions are a way to limit state power. In the past, an absolute monarch could dispose of anyone’s life or property. Criminal law, or human rights law, limits absolute power. The former-king-now-democratic-state must now comply with a fair trial, equality before the law, prohibition of torture, access to information, etc.

From a historical perspective, the ruler had absolute power, while individuals who were vulnerable and powerless organised themselves to limit the King’s or state’s power.

The 21st century is complicated by the fact that there are high net worth individuals, oligarchs and multinational companies that have far more power than countries: they have more capital, more media power, armies of enablers to escape paying taxes, can bankroll violent actors and bankrupt journalists investigating their affairs, and lobby or bribe legislatures to engage in rent-seeking.

Complete beneficial ownership information is about obtaining information about these powerful individuals who set up complex structures to escape the rule of law. It is a way to give more tools to authorities and to everyone else with access to beneficial ownership information, in the process helping to protect minorities and vulnerable individuals.

Even if there is mistrust on how authorities will use the collected beneficial ownership information, the beneficial ownership data will mostly refer to the (more powerful) individuals who are able to afford setting up companies and trusts, not vulnerable and discriminated populations on low incomes.

Some argue that some states are corrupt, or are dictatorships or autocracies, and that a beneficial ownership definition without thresholds would give them even more power to be abused. Unfortunately, those states are most likely already (legally) allowed to collect beneficial ownership information without applying thresholds, or to use other ways to obtain confidential information or coerce their citizens. 

Major financial centres where the rule of law is respected and where democracies do work, should start collecting complete beneficial ownership information. This way, democracies will be able to prevent oligarchs and dictators from creating entities and holding assets in these democratic financial centres, or will at least be aware of their interests and control, in case sanctions are to be enforced.

Securing data against hacks

Claim: All databases can be hacked. Collecting and centralising a trove of personal data on individuals related to legal vehicles creates a high hacking risk.

Our response: Although all systems could possibly be hacked, this has not stopped people from using banks, apps or password storage services. There are multiple privacy enhanced technologies that could be applied to reduce the risk of hacks or misuse, and to run analytics without sharing confidential information. 

For instance, the OECD report “Emerging Privacy Enhancing Technologies: Current regulatory and policy approaches” describes the use of:

  • Data obfuscation tools such as zero-knowledge proofs, differential privacy, synthetic data, and anonymisation and pseudonymisation tools which alter, create noise or remove identifying details.
  • Encrypted data processing tools, such as homomorphic encryption, multi-party computation, private set intersection and trusted execution environments, which allow data to remain encrypted while being used.
  • Federated and distributed analytics, which allow data to be pre-processed at the data source so that only the summary statistics or results are transferred.
  • Data accountability tools such as threshold secret sharing, and personal data stores.

The Bank for International Settlements Innovation Hub’s Nordic Centre in 2023 published a report on Project Aurora, a proof of concept that explored new ways of combating money laundering with a combination of payments data, privacy-enhancing technologies, artificial intelligence and enhanced cooperation across institutions and borders.

Ability to process data

Claim: Authorities are already overwhelmed with data and are often unable to process it effectively. There is no point in filing even more data with authorities because they won’t be able to use it.

Our response: The complete collection of beneficial ownership data is supposed to be handled by technology, algorithms, big data analysis, etc, and not necessarily through manual analysis. Complete data would create a deterrent effect, just as the implementation of automatic exchange of bank account information did.

The potential availability of complete data on all foreign bank accounts led to more tax revenues in many countries through voluntary disclosure programs (taxpayers voluntarily declared their foreign accounts and paid reduced fines, before the information was exchanged). Apart from the deterrent effect, even if data isn’t processed immediately, it could also be used in the future – many financial crimes have a statute of limitations of 5 or perhaps even 10 years, giving enough time for authorities to make use of it. 

At the same time, there is an exponential growth in technological capabilities around data storage, data matching and mining, and the processing of big data – all the while often becoming more cost-efficient. Just because an authority cannot afford data management functionalities today, does not mean they won’t be able to do it tomorrow. The prevalence of under-resourced authorities is not necessarily an argument against the collection of beneficial ownership information without thresholds, but rather an argument in favour of giving public access to information so that other stakeholders, including financial institutions, journalists, civil society organisations, researchers and foreign authorities can also use and process the data.

Conclusion and recommendations

An effective beneficial ownership framework would secure sufficient information to identify all individuals who may turn out to be relevant, after running advanced analytics to detect undeclared relationships and other red flags. Most of the proposals of this brief originate from the Tax Justice Network’s Roadmap to Effective Beneficial Ownership Transparency. This new report explains why these proposals are necessary. 

In a nutshell, the proposals include the following:

1. All legal vehicles should fall within the scope of registration – without exceptions

2. A “necessary” data approach should be applied, rather than a “small” data approach. This would ensure that all individuals who are related to a legal vehicle are identified whenever they:

a) Have control, ownership or derive benefits from a legal vehicle;

b) Have at least one share, vote, right to benefits, interests or exposure through financial instruments (ie without applying thresholds); or 

c) Have control via other means based on a non-exhaustive list such as power of attorney over the entity or its assets.

3. All “necessary” details should be required. In addition to basic identity details (name, address, nationality, date of birth), more details should be collected, such as: politically exposed person (PEP) status, tax residencies and nationalities (eg based on golden visas), identity of direct family members, price or value or reason for becoming a beneficial owner (eg price of transfer of shares), source of beneficial ownership (eg transfer of shares, apportionment as trust beneficiary.)

4. The full ownership chain should be obtained and verified, up to each individual with at least one share.

5. Charge fees for verification of complex structures. As a way to discourage complex ownership structures with too many layers and shareholders (which may increase costs for those collecting beneficial ownership information), financial institutions and beneficial ownership registries should be allowed to charge fees per layer and per shareholder for any entity that wishes to be incorporated or open a bank account.

6. Proper verification responsibilities. Countries should take an active role in beneficial ownership registration and verification by having the beneficial ownership register collect information on the full ownership chain down to each beneficial owner with at least one share, as well as cross-checking data and applying advanced analytics to detect red-flags (eg based on individuals’ declared income, income patterns for the neighborhood where the person is based, etc). This would reduce costs for financial institutions, which should be required to conduct some additional verification (not generally available to authorities) such as checking who administers the bank account, withdraws money from an ATM or from whom and to whom bank transfers are made.

Related articles

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.