2016-02-04 Meeting notes

DW: announcements. About done with entire slate of speakers for FIBO Days at EDW. MU and LC will give a talk on loans work there. Most of speakers are from companies that are council members. Will be sending out emails to the members, hoping to get more attendees.

Good meeting this morning with State Street and Dunn & Bradstreet (sic?) regarding a PoC. Some feedback for FIBO. Went very well.

LC: MU and I still working on roadmap, will come back to review this with group, after we review with other groups where there are dependencies.

JIRA Issues: Ticking them off, minimal discussion for the most part.

Inference demonstration: ConformingLoan defined to be a LoanContract that is guaranteed by either USDA, FHA or VA. The inference works, but it slows down inferencing dramatically. This is really a performance issue, not an ontology/logic issue. This is being considered by the FLT.

MG: want to define many such inferences, using the various properites

LC: ConformingLoan should be inferred to be MortgageLoan

LC: presenting small deck going over a number of properties that can be used to categorize loans, only some of them are such that the data is readily available. Specifically, what properties are such that enable one to infer that a loan is a HMDA-Covered Loan.

(see link to this deck on the meeting notes agenda page)

LC: commercial vs. consumer loans

MG: prefer idea of insured loan, not HMDA covered loaned.
LC: no, not in US. HMDA says certain kinds of loans need to be reported. Those are the one that are covered in the regulation.

MG: ok, a regulation. Properties indicate whether a loan should be included in the act.

LC: exactly.

MG: can use same properties to infer other things, not just HMDA, which is just US-specific.

LC: a number of the criteria/properties require data that is not readily available. Some of this information is just assumed, in its local context, would never be included in the loan documents/data.

LC: conundrum, the performance hit if we do this inference. Secondly, if want to infer HMDA coveredness, there would not be all the data you need.

LC: comment?

MG: have to be careful about assuming what organizations know about the loan they acquired. Might have a concept of ‘unknown’. Not want to infer yes, if there is an unknown, would need to have a flag to investigate.

LC: problem is, you would have to throw the flag on almost every loan.

MG: some organizations might be more sophisticated.

MU: what would motivate an org to do this?

MG: cost of making mistakes (fines). Cost of having to do the work manually. The ontology should allow it.

LC: use case was not to identify HDMA-covered, It was to support the information coming.

MU: maybe we should wait and do this later.

JC: can be taken as assumption of use case, that the organization knows which loans are covered an which are not.

LC: action, we will continue with the harmonization use case.

NOW: go to spreadsheet to go through more text definitions

LC: I see the mortgage as the note.

LC: The contract, the note. Also the security deed

Rename: CollateralAgreement to be SecurityAgreement.

TODO: change model to remove the subclass of SecurityAgreement called Mortgage.

MG: SecurityAgreement, generic to another group (e.g. FBC). Then collateral distinctions, real property (i.e. land) vs. personal property. May also be communal property, e.g. condos, associations.

MG: collateral, very complex law about this.

LC: lien positions not modeled in useful manner. Have it be in the singular.

LC: cannot easily discern the lien positions, might be primary, might know there is a secondary.

LC: impractical to expect a ordered list of lien holders, we do not always know who they are. Some have 1,2,3,4 others say primary and subordinate.

MU: can you sometimes know?

Can know position, if no default.

LC: from risk management perspective, want to know risk.

… looking at some other definitions, HMDA and MISMO.

LC: a title search is supposed to resolve this. Is complicated. At origination, they initiate a process: are they in primary position or secondary or lower?

MU: is it correct that we only care about whether the lender is in primary or secondary.

LC: yes, TO DO: remove class LienPositions, add property called lienPositionOfLender. Value is primary or non-primary. Will hang off of the SecurityAgreement.

MG: only relevant in some kinds of loan. Lender may not be able to guarantee their position.

LC: what is the relationship between a LoanContract and the SecurityAgreement?

LC: will not always be a security agreement, so not a necessary restrictions.

MU: what, in English, is the relationship between these things? Does it make sense to say: “The loan has a such agreement”?

Matt Barret: something allows a lien position to be updated, might be on a napkin.

LC: might have a lien that Is not a mortgage.

Perfected lien, vs. un perfected, latter means not legally binding, not official.

LC: The loan is secured. The security agreement is documented in a contract.

Copy of LoansAnnotations-2016-02-04_Team updates.xlsx

Decisions

Action items

Lynn Calahan [X] (Unlicensed) Jennifer Bond-Caswell (Unlicensed) Loans Roadmap
Michael ConformingLoan should be inferred to be MortgageLoan. DONE Rename: CollateralAgreement to be SecurityAgreement. DONE Remove class LienPositions, add property called lienPositionOfLender. Value is primary or non-primary. Will hang off of the SecurityAgreement.DONE