Dennis Wisnosky



Discussion items 20150630 FIBO-BE FCT.docx

20150630 FIBO-BE FCT


Walk­through of the taxonomy of kinds of legal or business entity.  Note that business entity as intended here is intended to be very broad, not a legal formal definition. The reason we created class called "Business Entity" was (1) because it's an intuitive concept and (2) because in banks on­boarding, they need to identify those entities which are on­boarded as business accounts.  That is, Business Entity is synonymous with potential business bank account customer in a retail context.  These are logical grouping of kinds of entity.  So for instance a sole proprietorship is considered a business entity, since it has a business bank account, as is an unincorporated association, a non profit, a club and so on, as these all get given business bank accounts as opposed to retail consumer accounts.


Legal Entity is a child of Business Entity and is all the things that are eligible for LEI because they are all potential market participants.  These include tribe, regional etc. "entities", that is the entity in terms of the holder of the bank account.  A governmental entity is considered a formal organization as well as being something that is eligible for a business account.


So far we don't have features that differentiate these.  Some of these are somewhat jurisdictionally defined. So this is the uber­view that includes everything.  We previously also looked at variants of those that are unique to the UK, US and so on, which were not in the initial release.  For this next release we will add jurisdiction specific ontologies or modules for US and UK, with a view to including additional country or state specific ontologies in future releases.


Governmental entities, in this definition, incudes things like Indian Tribal Entity (American Indian not Indian), Municipal Entity, Regional Government Entity and so on.  At present these are differentiated only by metadata in the definitions, but the intent is to add formal restrictions, with reference to OWL properties, to uniquely differentiate every sub­class in this hierarchy.  Investment Funds are also eligible for LEIs (see previous FIBO where these were a functional entity, and where Trusts explicitly incudes funds types of trust).


Mike note that investment funds always take some kind of form.  These may even include foundations, statutory bodies (see e.g. Kuwait Investment Office KIO).   So it makes sense for Investment Fund to be a functional Entity.  If something seems fuzzy, look for a verb. If the label has a verb it's probably a functional entity.


There is also a CIV section in FIBO Securities draft, which explicitly models all forms of fund, per EFAMA work (needing additional US and South African etc. input).


David recommends we remove Investment fund from the current hierarchy since the end result of looking at the existing models and of future work will be that it is no longer in this taxonomy of kind of independent thing as autonomous agent.  As long as we have the full range of things that could elect to register to function as an investment fund, then we are good. Whichever kind of thing registers to be an investment fund, needs to exist, and register, and be given an LEI. 


Action: David will move InvestmntFund to RelativeThing (under FunctionalEntity), but first will run this past Jeff Braswell in case there are any cases that fall outside of what we are saying here.


Question: Kristin: in getting an LEI – there is one scenario where this might not hold. We need to think about the current requirement to allocate LEIs to funds that that are trading swaps. Having thought about this some more, Kristin agrees that these are acting as legal entities. So to the question, ‘Do these fall into the definition we are describing?’  Yes, there must be some kind of entity (corporate entity, partnership or trust) that is already trading (in order to own or buy anything), so Kristen agreed that this is probably consistent with what we have been saying


Moving on: looking at things like limited liability companies, which are only defined in the US, while other forms may exist elsewhere.  Joint Stock Company is defined in the US as well.  Current model does not include the UK specific jurisdiction defined entity types.  Reloaded so we are also looking at that.  Unlimited Company is defined in the UK, and is something with unlimited shareholder liability.


In the US, a Joint Stock Company is one that has unlimited shareholder liability.

Designations e.g. PLC in the UK for Public Limited Company. hasLegalFormAbbreviation is the data type property that would apply these, and would have text like 'plc' for the above example.


Added new US concepts such as series limited liability company.


Trust: has added a number of sub classes to Trust, alongside the existing Trust and fund related trust that were there before. There is an almost open ended bit of these.  Trusts will need new restrictions to distinguish one from another.


Statutory Body, Supranational Entity are also eligible for LEI.  Chartered bodies and statutory bodies are universal concepts, and there will be unique names for these in different statuary jurisdictions, different kingdoms or principality.


The concept "Company" is not currently defined with OWL restrictions.  There is a legal perspective on this, with a most common definition, which typically means something that is a corporation, partnership ,business trust or foundation. We might also consider it appropriate to include sole proprietorship in this grouping.


So company is undefined now. It also has no children. It might be a logical union of those various things, or as a synonym for business.


Elisa: inclusion of governmental entities as being business entities. Not clear we would include governmental entity as a business entity?  Depends on what is meant by business entity?  Answer: the answer is that that it is something that a bank would on board as a business account.


Mike does not think that organizing all this around what can open a business account, is the best organizing principle.  Also these have "Business objective".  However it is not clear what is meant by "business objective".


Business has two meanings: business as in non technical, real world stuff, and business as in profit making.  What is relevant to the retail bank at Wells is the agency or whatever, that gets to participate in the market.  So these would be the market participating agencies of the government.  The things that have "business" constraints, maybe there is some union of legal entity and sole proprietorship and so on (maybe Company), then those things in that union are entities of the sort that have those kinds of characteristics. Would not include the chartered entity.  The label "Business” is a dangerous label because it has 2 meanings.


Our use case is much broader than just market participants, not just things that have

an LEI.  Not all things that have a "business" objective in one sense. For instance statutory bodies, legislative bodies and so on.  Our use case includes describing those entities that are Regulators, for example. Need these to describe regulations, laws, licenses and so on.  So that is much broader than either market participation OR LEI eligibility.  So we need a much higher level of kinds of Autonomous Agent, before we zoom down into what entities can open a bank account.  You need those representations to make the necessary distinctions in the regulatory space, which FBC are working on now, so we are already seeing the broader use case.


Elisa is very concerned about this.  Mike is very concerned about the chosen central organizing principles of the model as it stands now, which (unlike Beta1) does not frame things in terms of the legal realities of the types of entities themselves.


David: So, for instance the things they have defined as e.g. Government are the things that can open a bank account.


We can see that we need the kinds of things in David's hierarchy, but Elisa is not happy to see these under a wrapper called Business Entity.


Mike considers that Government Entity is trying to conflate 3 kinds of things, which include relative entities, legal persons including sovereigns), and those physical organisms which work to discharge those responsibilities.


David needs to hear more feedback form the legal folks, including Wells, on what will be useful for the banks.


Steve Creek and others (maybe Mike Pool) and Gareth Issaac, can all provide us useful feedback on this area. We should run the current proposed taxonomy by them.


Even if GovernmentEntity is something that can issue bonds, Elisa feels that

classifying them as Business Entity is still a step too far.  Please post the slides from this taxonomy, that folks in the other FCTs can review, with the formal definition of BusinessEntity, LegalEntity and so on.  David: these are poste on the BE wiki at present.


Similarly there are things in FBC which the BE and other folks e.g. Loan need to validate and vet.  David has concerns about possible constraints in FCB. Would prefer to see some common forum rather than needing to come to FBC.


Dennis: the FLT is already supposed to do this.  However, consider that in Elisa's case there are many knowledgeable folks form the banks, whereas the FLT is a smaller group.  So that would not be a viable solution either.  In the FLT as we have been focused on the testing to ensure that the ontology works.  This must take precedent over opinion.  mNevertheless because this seeming conflict between FBC and BE definitions is not going away so:


Action:   Dennis will call a special meeting to address this purpose.  This will focus on how we classify the concepts, rather than the automated testing of the physical integrity of the resulting models.  At present there is clearly ambiguity in the BE models, so it makes sense to meet with the FBC folks and get their perspectives.


Ownership and control.  This is an example of the Lattice Pattern.  There is e.g. an Ownership lattice. (Note: Lattice is the low level pattern; Partitions are the top level pieces of the upper ontology e.g. Relative Thing).  The lattice describe how the partitions are to be applied in a specific business concept area, such as ownership, control, etc.  The relative to independent relation is called isPlayedBy,with the inverse being  playsRole.  The ownership lattice is symmetrical ­ there are owners and assets (this is not the case for all uses of the partitions).  This gives two roles, as perspectives on the instances of something playing that role in that context.  There are direct relationships of ownership between independent things.


For sets of concepts where the upper ontology partitions are used differently, there are different ontology design patterns.  It is not clear whether we are reserving the word Lattice just for the symmetrical ownership and control patterns or for all such patterns.  We need to converge on a clear terminology about this.


Action:  FLT will work on the set of patterns that we want to set as standards for FIBO.  And, whether to call them different kinds of Lattice, or whether to reserve the word  Lattice for ownership and control exclusively.


In Protege, we look at Ownership.  This is defined in Foundations.  There is an owned thing and an owning party. Ownership is a Mediating thing.  There is a set of sub classes of Ownership, based on kinds of owning interest.  These include controlling interest (which combine=s control as well as ownership).  We should regard Dean's initial set of sub classes of Ownership, as a straw man.  The names may also change.


AnyInterest: the most generic form of ownership (each of these classes represents one more instantiation of the Lattice pattern, with specific labels for the relationships between the boxes in the Lattice slide, e.g. Interest Holder, hasInterestHolder and so on, for AnyInterest, ControllingInterest etc. etc.  David added TotalInterst, which represents 100% ownership.


Discussion:  Do we want all these sub classes of Ownership, do we want to revise some of them, are they at the right level of specificity.


Q: Majority interest versus Controlling Interest?  A: Majority is usually 50% + 1.  Controlling interest is very generic ­ not clear.  Thompson Reuters uses Any Interest and Controlling Interest interchangeably.  The TR model has Controlling Interest as anything that is going to force that parent relationships, similar to the 50% +1; other ways to deem something a parent relation.  Then there are also influence relationships, which are typically 50% or less, but has some significant relations.  Then even within that, you might be "deeming" such based on a cut­off of 20% or 25% or other borders to this term. These are all affiliate relations.  Then there is the Joint Venture relationship, which may also be control, majority or other. Can be duplicative.  So there are nuances even in the simplistic breaks that TR chooses.  Recognize that there can be a controlling entity, aggregate ownership, indirect ownership and so on.


MB explains the original FIBO distinction between controlling interests and de facto control.


Kristin: that makes sense.  Here we are taking about control, on the assumption of ownership that also comes with control, but in reality there are non voting shares and other forms of non controlling equity.  Also there are side contracts, so you can have 100% ownership and no control, or cases where you have 100% ownership and 100% control ­ these are different concepts even when (as usually the case) the

implications normally follow.  Similarly ownership does not imply (does not mean) control) but there is control which  follows from ownership.  And control which doesn't, e.g. side contracts, which do not endow ownership on the counterparty.  TR and other folks follow the idea of "deemed" subsidiary, parent etc. relationships based on lower thresholds.


Mike suggests this may correspond to cautious reporting requirements from central banks, in that they want to see the broader levels of ownership so they can dig deeper into whether that ownership confers control or not.


Corporate Control is also there, with a set of SWRL rules for test.   Will run Pellet (a reasoner) on some test cases).  This will classify instances, showing assertions about who holds what interest in who, and who owns whom generically and so on.  Includes results about when something holds a total controlling interest in some

other bank.


Need to walk through what kinds of inferences we want.  Significant controlling interest is between some threshold and 50%, then majority is 50% +1.  David presents two possible treatments for these:  Whether to nest these or treat them as separate kinds of interest.  That is, should a query into all the significant controlling interests in a thing, also include the 50% + 1 and the total interests. Need to know if that is right i.e. are people looking for the most specific case or the least specific case.


There may not be one right answer to this. Need people's input to this.  If these are made into independent assertions, then the queries will be easier, but a query into significant interest, will not return majority and total ownership relations in the same query result, which could be misleading.


David suggests there is not a single right answer but we need an answer.


Action items Dave Newman David recommends we remove Investment fund from the current hierarchy since the end result of looking at the existing models and of future work will be that it is no longer in this taxonomy of kind of independent thing as autonomous agent.  As long as we have the full range of things that could elect to register to function as an investment fund, then we are good. Whichever kind of thing registers to be an investment fund, needs to exist, and register, and be given an LEI.  David will move InvestmntFund to RelativeThing (under FunctionalEntity), but first will run this past Jeff Braswell in case there are any cases which fall outside of what we are saying here. BE-47 - Getting issue details... STATUS

Dennis Wisnosky Call a combined BE and FCB meeting.  This will focus on how we classify the concepts, rather than the automated testing of the physical integrity of the resulting models BE-46 - Getting issue details... STATUS FBC-11 - Getting issue details... STATUS .