Special Working Group LGD Reference Model (15/7/2019)

Dear LGD reference model WG members,

Since two weeks, we’ve had discussions with FCG, Pubudu and the chairs of the WG.

The situation is now clarified in that the extension would require a delay of minimum 6 weeks with risks regarding the validity of applying the LC RDS onto this new scope.

Therefore, it is asked to vote pro or cons the extension of the scope.

Please vote by email by replying to this email to olivier.plaetevoet@globalcreditdata.org or by attending the last special WG this Monday July 15th at 2pm-2:30pm CET / 8am-8:30am EDT / 10pm-10:30pm AEST  .

Detailed information:

Arguments for the rescoping:

a) The current study is based on the assumption that the Large Corporate segmentation is representative for LGD modelling. Basing on this assumption may lead to inaccurate results.

b) The study is not relevant for banks that don’t use the Basel FAC segmentation (eg. Canadian banks).

Arguments against the rescoping:

a) An estimated delay of minimum 6 weeks will occur given that we need to rescope from a LC to a total scope. It means Modelling done by mid-September and Documentation done by mid-October. This planning doesn’t take into consideration the definition of a new RDS! Therefore, FCG estimates there is a strong risk that the conclusion will be incorrect in case the RDS doesn’t fit the larger scope (see points below).

b) The Large Corporate scope was chosen because we decided to start the WG in a controlled environment (with a validated LC report and RDS). Once the method proves to be validated by satisfying results, we can direct the study toward a larger scope in a second stage.

c) The purpose of using a RDS conflicts with extending it to a different scope because it is defined for a specific representativeness purpose :

“The purpose of creating an RDS is to match as closely as possible the risk conditions of a target portfolio of a single bank.  Therefore, a single standard RDS could not possibly suit all users.  In this report GCD bases the analytics on a filtered data set which combines elements of representativeness and data quality.”

d) Therefore, if we don’t take into account point (c) and still enlarge the currently defined RDS for LC, it will probably not fit the whole portfolio; for the following reasons:

1. For the Large Corporate RDS, member decided to exclude borrowers with an EAD less than 100.000 euros. Can this threshold be meaningful for the entire data set? Even for SME?
2. Members also decided to exclude borrowers with more than 10 loans, expecting that this might cause overweighting. Is this applicable to Project Finance asset class for example?

e) Creating a new RDS on the whole portfolio is a time-consuming exercise. Based on the previous experience of defining the LC RDS, it took more than a year including validations.

Please submit your vote (for or against the rescoping) by replying to this email (to olivier.plaetevoet@globalcreditdata.org ) before the special WG. If you wish to discuss it, feel free to join the working group this Monday 15^th July between 2pm and 2:30pm CET / 8am-8:30am EDT / 10pm-10:30pm AEST .

Aganda:

https://www.globalcreditdata.org/wp-content/uploads/2019/07/wg_lgd_reference_model.ics