Credit Risk+ CSFB model

[Sidharth]

Hi

My doubt is on Credit Risk+. I was going through the CFSB doc at following link, In the first model where we come across a recursion to create distribution. (equation 25 page 40)

At that point we need data input for E[j] and V[j]
where j is 1<j<m where m is the number of exposure bands you have created
E[j] - is the expected loss in that band
V[j] is exposure in that band.

I want to create loss distribution of portfolio having two bonds worth 50mn and 100mn, I will take minimum unit of L to be 5mn and i create 3 following bands

0-25 mn first band V[1] = 10 (25 from first bond and 25 from second)
25-60mn second band V[2] = 12 (25 from first and 35 from second)
60-100mn third band V[3] = 8 (40 from second)

Values of E[j] are not very clear to me , assume both bonds depend on same risk factor with probability of 2% default and have recovery rate of 60%.

Is it okay to assume E[j] can be = 2% X V[j] hence
E[1] = .2 E[2] = .24 and E[3] = .16

Taking above into count , I am not able to get all the values for the loss distribution , I have attached spreadsheet i used, can you please take a look and let me know if anything can be improved here.

Regards
Sidharth

Link
http://www.csfb.com/institutional/research/assets/creditrisk.pdf

 

[AG]

Hi,

I recently came across this model... I dont know if you still require clarification on this or got it resolved by yourself...

First of all L automatically defines the bands... If L is 5 million, the bands will be of width 5 million.
you have two exposures in this case. v[1]=50/5=10, v[2]=100/5=20; accordingly E1=10*2% and E2=20*2%.

That's it. this is the input to a basic CR+. You will get the loss distro in units of 5 mn. that means y'll have to scale by 5mn.