Credit PortfolioView Model

[liewpw05]

Hi David,

I have two questions.

1. Is the Credit PortfolioView model described in Desevigny Chapter 6 (in Core Reading Year 2008) the same as the McKinsey & Co/Wilson model in Chapter 20 (in Core Reading 2009)?

2. From Desevigny Chapter 6 the description of the model stops at generating a distribution of migration probabilities. How is the portfolio loss distribution generated from there? Does the steps in Stimulation Framework Figure 6-2 applies?In particular the stimulation of correlated returns is still performed and from there the use of the generated transition probabilities to come up with the generated migration events?Then using an asset valuation model to value terminal portfolio values and hence portfolio loss distributions.

Thanks

Peggy

 

[liewpw05]

Hi again,

Just to clarify the McKinsey & Co/Wilson model i am refering to is from Caouette, Chapter 20.(Core Reading 2009)


Peggy

 

[David Harper CFA FRM]

Hi Peggy,

Yes, the model is definitely the same. I'd imagine there is some reason Caouttee Altman avoided using the McKinsey "branded" name of CreditPortfolioView, but they do title it "McKinsey & Co/Wison" and that is clearly the CreditPortfolioView (CPV) in de Servigny.

As to the detail, I don't really know have insight into CPV beyond the de Servigny assignments (sorry). It *seems* to me that Fig 6-2 would be inadequate to the task, as it refers to CreditMetrics and while they are both migration-based approaches, my take is that CM gets there via the correlated asset returns (CM extends Merton model and is therefore inherently *structural*) but CPV generates conditional transition matrices based on simulating economic variables...sorry i can be more helpful on the specific here...David