Simplify the entity type codelist

[ScatteredInk]

We currently have five types of entity: registeredEntity, legalEntity, arrangement, anonymousEntity, unknownEntity.

registeredEntity and legalEntity are a source of confusion due to conflicts with their common usage in other related standards, for example most legal persons under FATF guidance would be referred to as registered entities in BODS.

I suggest removing the registeredEntity category and aligning "legal" and "arrangement" categories (broadly) with the definitions used by FATF for legal persons (R24) and legal arrangements (R25).

legalEntity (or legalPerson) would then cover registered companies and the cases currently covered by legalEntity. Any additional details about the entity (for example, the fact that it is a government department with legal personality) can be covered through the proposed entity tagging system (currently #186 but full issue to follow) or by the use of identifiers. One advantage of using the -person rather than -entity postfix here is that our statements would become more generic, allowing us to add an asset type if we wanted to at a later date, without it looking too odd.

@siwhitehouse As currently stated arrangement would cover the legal arrangements as described by FATF Rec 25 (which are broadly speaking express trusts and trust-like arrangements) and any legal arrangements that give rise to ownership and control interests (like a revenue-sharing agreement or nominee shareholder agreement). We may want a separate term for these to allow for an easier distinction.

(Note that other entries on this list may change due to work @rhiaro is doing on structuring missing info.)

[kd-ods]

I think this proposal will work.

At the outer edges of the 'legal person' FATF definition are legal persons which are not - in our terms - registered entities. I'm thinking of unregistered clubs, societies and associations in the UK context, for example. If we just have 'legal entity' and 'arrangement' types then we'll need to be clear in the documentation under which of the two categories such unregistered bodies sit.

On the question of terminology - 'legal person' or 'legal entity'. Ooof. People with English as a first language can find the concept of natural vs legal personhood a challenge, so I think we should avoid it if we possibly can. BUT I do see why we need to consider it. Difficult one.

[ScatteredInk]

Proposal

Having thought about this a bit more, I would propose the following types:

Legal entities

This would cover legal entities as defined by FATF:

Legal persons refers to any entities other than natural persons that can establish a permanent customer relationship with a financial institution or otherwise own property. This can include companies, bodies corporate, foundations, anstalt, partnerships, or associations and other relevantly similar entities.

The LEI ISO standard 1744201:2020 has a similarly expansive definition, based on abilities/actions rather than legal forms:

[legal entities] include, but are not limited to, unique parties that are legally or financially responsible for the performance of financial transactions or have the legal right in their jurisdiction to enter independently into legal contracts, regardless of whether they are incorporated or constituted in some other way (e.g. trust, partnership, contractual). It includes governmental organizations, supranationals and individuals when acting in a business capacity

@kd-ods I think this covers the edge case you are referring to quite nicely - so we are now less worried about the legal structure than the actions taken by the legal structure.

Examples of a legal entity would include: an incorporated company, a government department with legal personality, a government department without legal personality able to enter into a relevant custodial or contractual relationship over an asset.

Legal arrangement or trust

I suggest we restrict this type solely to trusts and trust-like arrangement, as defined by FATF:

Legal arrangements refers to express trusts or other similar legal arrangements. Examples of other similar arrangements (for AML/CFT purposes) include fiducie, treuhand and fideicomiso.

Trusts are modelled in a predictable way (based on previous work to build compatibility with CRS) and have a different risk profile from both companies and agreements. We are starting some research that may lead to further changes in 0.4 but it makes sense to retain a separation here, given that there is an explicit separation in the dominant international standard driving record-keeping and disclosure.

Agreements

An agreement is a formal or informal nexus through which ownership or control interests are brokered between the relevant parties. We currently treat anything that is not an entity as equivalent for modelling purposes, although a trust is very different from an informal agreement to share profit or a joint-shareholding agreement. These methods of brokering control are a key risk area and we want to be able to properly model and monitor their use, especially as more BOT regimes are put in place and avoidance efforts become more sophisticated.

I would expect the agreements type to be very diverse and, over time, even point towards new entity type. But for now we could simply use tags (bearerShare,...) and a description to get the details.

Examples: a bearer share agreement, a profit-sharing agreement, an oral agreement to pass on the all profits received as dividends, a nominee shareholding.

(This concept was raised previously in #92 but tied into some structural changes that aren't really necessary.)

User stories

1. as an analyst I want to understand systemic patterns of use of particular legal entity types so that I can understand how ownership or control is changing as a result of a new disclosure regime
2. as a systems designer I want to understand how far I am conforming with existing international disclosure standards so that I can improve our technical- and outcome-based assessments.
3. as an analyst I want to distinguish between "trusts" and other legal constructs because these have different risk profiles and modelling needs so that I can spot the gaps in a disclosure, verification and enforcement regime
4. as a register analyst I want to be able to add suspected data on informal agreements between parties to disclosures so that I can flag disclosures as suspicious and pass enhanced disclosures to the relevant authorities
5. as a publisher of enhanced data I want to build on existing disclosures and include informal agreements so that I can highlight inaccuracies in BO data and ask for improvements.
6. as a policy researcher I want to be able to explain how BODS conforms to existing domain practices and enables high-quality data publication as a means of meeting international disclosure standards so that I can make BOT reforms more effective

User story assessment

Much of the risk assessment work will involve good identifiers and for identifiers to map to entity types. That won't always be the case so:

• we can create guidance on using annotations on entityType to put in local descriptors (I don't really like this and long term think we need a better method for localisation).
• we could potentially have sets of tags that applied to legal entities, arrangements and agreements if we think that these are going to be disjoint (or make them open codelists if we're not confident). This would avoid a later untangling job and the potential for a legal entity being tagged as 'bearer-share'.

[kd-ods]

... so 'legal arrangement' would be very narrow in scope, but 'agreement' would be extremely wide-ranging in scope? That's OK, I guess. Though, I do wonder whether we should allow legal arrangement to extend beyond the minimal FATF scope, but tie the FATF definition (express trusts etc.) to a particular arrangement tag. That would allow for extensibility and inclusion of other very 'standard' legal arrangements in future. For example, particular joint venture arrangements.

we could potentially have sets of tags that applied to legal entities, arrangements and agreements if we think that these are going to be disjoint (or make them open codelists if we're not confident). This would avoid a later untangling job and the potential for a legal entity being tagged as 'bearer-share'.

The other advantage of going down this route from the start is that the structure will help people with mapping their information onto the data standard: making it easier to understand what kinds of things are agreements vs legal arrangements, for example.

[siwhitehouse]

I think it's wise to separate "legal arrangement" and 'agreement' so they better reflect how FATF categorise them.

One thing that has been preying on my mind since looking into bearer shares are the different ways of categorising these entity types. The entity type titles - and our discussion - are very much aimed at how the entities are constituted, how formal they are etc.. As a primary classification it makes sense and I'm not suggesting we do anything different.

For an entity or a trust it is arguable that they exist to allow a person, a state etc. to do things they could not do alone. So, they provide a benefit to their owners.

With the agreement type it is not so clear whether they exist to provide a benefit (informal agreement to share profit), exercise control (nominee shareholding) or provide a means of tracking/registration (bearer shares and custodians). I think that a variety of users are going to be interested in understanding the type of agreement based on this sort of categorisation. At the moment they have to work this out by having an understanding of the different types and - potentially - by reading the descriptions.

Having said all that, I'd like to see us develop the idea of the tags a little bit more and implement those before working on the sort of secondary categorisation I'm thinking about here.

[ScatteredInk]

Noting that in the proposal for state-owned enterprises we suggested two other entity types:

• stateEntity to represent a nation state that ultimately controls an entity;
• stateBody to represent a government body that owns or controls an entity, either directly or indirectly. (We might to think about international and multilateral bodies too.)

State entities are entirely new - we could previously only represent these artificially using an arrangement.

State bodies fit awkwardly across all existing categories. Thinking about data quality and analysis, the primary point of interest is that they are state bodies so my instinct is that we want this as the top level classification and that the question of registration / legal personality is secondary and can be picked up in the entity details if it is relevant.

[ScatteredInk]

From discussion in #339:

Add the expanded list of entity types: state, stateBody, legalEntity, arrangement, agreement, unknown, anonymous.

[kd-ods]

Bringing this ticket up-to-date with what is in BODS 0.3...

We now have an entityType and an entitySubtype property. And the former has values: registeredEntity, legalEntity, arrangement, anonymousEntity, unknownEntity, state, stateBody.

Looking at the original thinking on this thread, I agree that we should align things with FATF categorisation. That would mean, I think:

• removing the 'registeredEntity' type
• renaming the 'legalEntity' type to 'legalPerson'
• NOT adding an agreement type, but...
• Adding entitySubtypes: 'expressTrust', 'jointShareholding', 'otherLegalContract' plus 'nominee shareholding', 'nominee directorship' and anything else that comes out of #329 and #328. (ie we use entitySubtypes to capture what we've called agreements on this thread)

[kd-ods]

@oalannao - in the coverage briefing we essentially say that all corporate vehicles are either 'legal entities' or 'legal arrangements'. We are likely to make changes to BODS which align with this 2-type classification (see above). Do you think that we should stick with 'legal entity' or should we move to 'legal person' like FATF?

[kd-ods]

From @tymonk :

Then the main distinguishing feature would be distinct legal personality or not? In some countries trusts have legal personalities

Actually, I'm rowing back on my earlier suggestion. Looking at @ScatteredInk's suggestion again, I think sticking with 'legal entity' as a term is a better idea.

But your observation still stands, @tymonk: legal personality would still mean that trusts could be represented in BODS as 'legal entities' in some jurisdictions. I think there are two issues here:

1) how should trusts be represented in BODS data (#328), and

2) should legal personality be the main distinguishing feature between those things in the 'legal entity' bucket and those things in the 'arrangement' bucket.

On (2), using 'legal entity' instead of 'legal person' helps us keep the line that: those types of entity that usually have legal personhood should always be represented as legal entities; and other entities able to preform financial transactions or enter into contracts should also be represented as legal entities. Usually, arrangements have no legal personhood.

On (1), I think BODS has to take a line on how trusts and trust-like arrangements should be represented. Given that in most jurisdictions they fall into the category of legal arrangement, I think that's how we will advise they are represented, regardless of their actual status in local law. But it won't be mandated in the standard: where are there is good reason, publishers can model their trust-like entities in another way.

I think the main point on (1) is that we shouldn't expect that it is BODS data that would - for example - help a researcher answer the question: globally, what proportion of jurisdictions grant trusts legal person status? It's more important that trusts are modelled consistently within data from a given jurisdiction than within data from different jurisdictions.

[oalannao]

@kd-ods thanks for checking on this. I agree that the Coverage briefing would support having 'legal entity' more than 'legal person' because entities don't always have legal personality, and as I understand it, if one does then that would be reflected in the 'entitySubtype' property? I don't think that legal personality would be the most appropriate distinguishing feature between entities and arrangements given the variations that can exist in both.

In the briefing, we actually pulled back on saying "legal" before both entities and arrangements to broaden it futher since both may or may not have legal personality. Parnerships an entity type, for example, that is often regristrable and may or may not have legal personality (maybe these would fall under agreements in BODS though?). So using simply 'Entity' and 'Arrangement' might be an option, but I think using 'legalEntity' and 'Arragement' gets close to that.

Regarding your questions

1. I agree with your reasoning of having them fall under arrangements as the default. Is there the option of using subtypes for 'arrangements' as well as entities that capture characteristics like 'isTrust' 'hasLegalpersonality' 'isTrustlike' (that one may be too subjective...)?
2. I agree with your assessment.

[oalannao]

@kd-ods looking again at tbe Box on page 7 of the briefing, I am reminded that one difficulty we had was finding an actual definition of arrangements, which a reviewer (lawyer) eventually provided: "Parties can establish legal arrangements to govern their relationship in pursuit of a common purpose or to create rights and obligations with respect to specified assets"

When you read it that way, it sounds a lot like an agreement, so I think not havnig agreements, as you suggest, will probably be helpful. Then partnerships should probably always go into entities despite not always having legal personality and being an agreement of sorts.

Also I just realised that the suggestion is there for subtypes e.g. 'expressTrust" so on the same page :)

[tymonk]

Legal entity has differing definitions but many use a legal personality when defining. FATF defines legal person as separate legal personality as well (although natural persons are excluded, while in law they would also be considered legal persons). Thinking about this more (and rereading Jack’s suggestion again) I think separate legal personality is the right distinction. This does mean though that there could be trusts from certain jurisdictions that would need to be “entities” in BODS. If this will not be too confusing I think this makes sense. If a trust has a separate legal personality and assets are in the name of the trust and not the trustee they are materially different anyway, as they don’t have the same “ownership limbo” issue as traditional trusts. Trustees will maintain their fiduciary obligations though.

On Fri, 9 Jun 2023 at 15:47 oalannao @.***> wrote:

@kd-ods https://github.com/kd-ods looking again at tbe Box on page 7 of the briefing, I am reminded that one difficulty we had was finding an actual definition of arrangements, which a reviewer (lawyer) eventually provided: "Parties can establish legal arrangements to govern their relationship in pursuit of a common purpose or to create rights and obligations with respect to specified assets"

When you read it that way, it sounds a lot like an agreement, so I think not havnig agreements, as you suggest, will probably be helpful. Then partnerships should probably always go into entities despite not always having legal personality and being an agreement of sorts.

Also I just realised that the suggestion is there for subtypes e.g. 'expressTrust" so on the same page :)

— Reply to this email directly, view it on GitHub https://github.com/openownership/data-standard/issues/336#issuecomment-1584704350, or unsubscribe https://github.com/notifications/unsubscribe-auth/AS327HGFX5XQO2VKKS76RZLXKMZOTANCNFSM4XLDFDQA . You are receiving this because you were mentioned.Message ID: @.***>

[kd-ods]

@oalannao @tymonk - Huge thanks for your reckons on this. To summarise, I think when this moves ahead we'll be looking at:

1) Removing registeredEntity from the entityType list, taking us down to four types of entity: legalEntity, arrangement, anonymousEntity, unknownEntity.

2) Using the entitySubtype property for further classification of arrangements.

3) Investigating what universal sub-categorisation of legalEnties will be useful, which can be added to the entitySubtype codelist.

[kathryn-ods]

Initial assessment of the impact of making this change for 0.4 - @kd-ods I think this is achievable for 0.4 but happy to take your lead on whether this is a priority.

Schema Removal of ‘registeredEntity’ from entityType codelist. Update definition of ‘legalEntity’ to be inclusive of current registeredEntity

Docs All the docs will need to be reviewed and any reference to ‘registeredEntity’ removed. Diagrams and alt text will also need to be updated.

In particular: SOEs https://standard.openownership.org/en/latest/schema/guidance/repr-state-owned-enterprises.html Upcoming trusts guidance Example data https://standard.openownership.org/en/latest/examples/index.html?highlight=registeredEntity

Test data Data use for tests will need to be updated where registeredEntity has been used as a data type - https://github.com/openownership/data-standard/tree/main/tests/data

Also draft 0.4 test data https://drive.google.com/drive/folders/1ezezz_8K7zDuwodDwhRdr-w1LR7BTfxs https://drive.google.com/drive/folders/1BQoQD4uYbDF-XISOsrdL_svxwTEqRpoN

New ‘invalid’ test data where ‘registeredEntity’ is used with message ‘registeredEntity is not a valid entityType. See the entityType codelist.’ or similar

[kathryn-ods]

Decision to put this off until next release after 0.4 to allow for consultation with users

[kd-ods]

Comment on #339:

There is an ISO standard (20275) for Entity Legal Forms. This list will be useful when considering subtypes of entities and how general and local categorisation might work. (It might be overkill to build ISO 20275 code usage into the BODS schema, but users' use of ISO 20275 values and our use of ISO 20275 terminology should be considered.)