Hi All,
Is there a way to simulate LGD dataset given the lack of LGD datasets in public domain. The variables can be LGD, EAD, Recovery cost, Value at Recovery.
Thanks,
Srihari
You may find my answer on a similar question helpful:
With LGD estimates you have two choices. One is you can use historical loan behaviour data, which is really only available if you work at a bank. Second is you can construct a model that simulates loan behaviour (you need to model PD as well to get LGD) but is also subject to econometric factors like house prices, interest rates, inflation rates, unemployment rates and so on. This is harder to do by far than option one. Even if you want to build a model just for practice it is tricky. Also the only way to calibrate your model is to compare it to real losses. Such models tend to be built by a bunch of stats and BI gurus hired by banks at great expense.
I'm not trying to put you off just trying to give you a realistic answer...
Hi Srihari,
Is there a particular SAS Risk Management product you are using that you need sample data for?
Regards,
Brendan
Hi Brendan,
No I just want to practice measuring LGD. I'm using On Demand Academics. I'm not able to find LGD data for obvious reasons so I thought simulation is a way to go(I'm not skilled enough to generate simulated data set at least for now!).
Regards,
Srihari
You may find my answer on a similar question helpful:
With LGD estimates you have two choices. One is you can use historical loan behaviour data, which is really only available if you work at a bank. Second is you can construct a model that simulates loan behaviour (you need to model PD as well to get LGD) but is also subject to econometric factors like house prices, interest rates, inflation rates, unemployment rates and so on. This is harder to do by far than option one. Even if you want to build a model just for practice it is tricky. Also the only way to calibrate your model is to compare it to real losses. Such models tend to be built by a bunch of stats and BI gurus hired by banks at great expense.
I'm not trying to put you off just trying to give you a realistic answer...
Hi, Thanks for the reply. I can see where you are coming from in terms of your answer. Most of the macroeconomic variables if not all tend to be not significant when it comes to measuring LGD. So I guess we can build a dataset sans macroeconomic variables at least for practice sake(I know its not ideal way to learn). I saw in SAS studio there is a procedure to simulate Regression dataset so I guess there must be a way forward to simulate LGD dataset with variables like LGD, EAD, Recovery cost, Value at Recovery.
Regards,
Srihari
Join us for SAS Innovate 2025, our biggest and most exciting global event of the year, in Orlando, FL, from May 6-9. Sign up by March 14 for just $795.