Hi there, I am junnior in this field. Need some help and clarification from our expert here .My bank is implementing BASEL II now and we are building an application rating/credit scoring and PD model at the same time. I believe from credit scoring we can derive the score/rating and PD and from PD model we would come out another model and the output of PD. So now we have rating with PD and addition PD from PD model. How do we map it? A bit confuse. I just wondering whether we can just based on the PD from PD model to treat as capital requirements inputs. Any views?

I'm not sure if this is what you're looking for, but my understanding is that you take the PD estimates and feed them through the risk-weight functions in Basel II to get the capital requirements figure. Now, there are all sorts of requirements for your PD-rating apparatus that you have to meet before you can implement Basel II for real. Are you at a US bank?

tks for your points. What i am saying here is not to feed the PD for BASEL capital calculation. I am concern about the alignment of application rating and PD from my PD models. You different rating outcome from different rating model/scorecard e.g. Large Corporation Rating (with PD if you use logistic regression)....then you also have the PD from PD models for this segment or corporation based on reduced form models (also logistics)...So how do you map it? Another question arise when you have so many different scorecards for different segments how do you align it before you map it to PD from your PD models?I am not from US bank. Ha US banks do not care about BASEL II IRB haha..

I am having great difficulty in determining how to use logistic regression and MDA to calculate PD. Can you help me please do you know of any paper also which software did you use for that

I am having great difficulty in determining how to use logistic regression and MDA to calculate PD. Can you help me please do you know of any paper also which software did you use for that

Hi,It's a bit old but fruitful discussion to me. I am busy with similar subjects novadays. In order to model credit scoring of SMEs i follow the path of "apply PCA to data, in order to find out best fitting financial ratios for a default case" and "apply Altman's z score model to the output".I read the paper on RiskCalc v3.1 of Moody's KMV, which is not an open source one. Is there anyone who tried to replicate KMV's or any probit/logit model on credit scoring?I will be appreciated with any repsonses, and thanks in advance,R