Value at risk with monte carlo simulation for credit risk

[COTBL]

Hi Guys,

Can anyone help with a program that calculates the potentiel future exposure for a given portfolio at specified period of time calculated at some level of confidence using monte carlo simulation?

Best Regards

COTBL

[Rick_SAS]

Sure. Post your program and use comments to highlight the sections that you have questions about.

[AndrePoorman]

Hi,

If you are willing to do some conversion, I can likely provide code in R, MatLab or Python.

I haven't come across any PFE (Potential Future Exposure) credit risk model in SAS so far.

Andre

[Rick_SAS]

SAS users have been estimating value-at-risk in SAS for decades. A Google search for 

SAS "value-at-rick" "VaR" simulation

will show you some conference papers. The first hit is an IML-related paper: https://support.sas.com/resources/papers/proceedings/proceedings/sugi29/200-29.pdf

and a follow-up the next year https://support.sas.com/resources/papers/proceedings/proceedings/sugi30/173-30.pdf

A more modern paper is Joshi (2018), which uses PROC COPULA and econometrics routines. 

Since the early 2000s, most banks have chosen to use a SAS Risk Management solution,

which provides an interface to creating simulations that estimate VaR by using historical or hypothetical market events.

However, you can certainly run the simulation yourself, if that is what you prefer. I offered the OP assistance but never got a response.

[AndrePoorman]

Many thanks to Rick_SAS which answers in large part my question.
The metric I am actually interested is Potential Future Exposure (PFE) which is in some ways a variation of VAR analysis for financial Credit Risk evaluation.
I think I can adapt some of the code indicated by Rick_SAS to my purpose.
Or is somebody aware of SAS code directly related to PFE in the community?

[AndrePoorman]

Here is a definition of PFE: https://en.m.wikipedia.org/wiki/Potential_future_exposure

[AndrePoorman]

Also some sample code in Python (I will look into using SASPy but direct SAS would perform better and be easier for us to maintain. https://www.google.com/amp/s/ipythonquant.wordpress.com/2015/04/08/expected-exposure-and-pfe-simulat...