PD model in Credit Risk Analytics

analytics
risk

#1

Hi,

Please someone give me a brief about PD model in Risk Analytics.
Also it would be helpful if you can post any link or document on how modeling is perfomed on credit score cards.


#2

@dada_kishore

This is a very broad topic to give a brief about!

Here is an attempt to do it:

PD model aims to predict the probability of default for the person / entity taking the loan / credit. The model can differ vastly depending on whether it is an acquisition stage model (where you only have variables available on the application and possibly bureau data) or a customer management stage (where you have information about customer behaviour as well).

It also depends on whether the credit is secured lending (like home loan) or unsecured (like creadit card or personal loan). Depending on the context, the significant variables and the models can vary. In the simplest form, the model could be a logistic regression. In a more evolved setup, it could even use techniques like survival analysis and Markov chains.

Typically, once the model is build, it is continuously monitored for input and output variance, as the sensitivity for wrong outcome can be huge.

Hope this helps
Kunal