CDS pricing models used in Bloomberg

seanmok · January 7th, 2008, 10:08 pm

Hi everybody,I have seen four models used in Bloomberg (page: CDSW) for CDS pricing:1. JP Morgan2. Bloomberg3. discount spread4. Modified Hull-WhiteCan you please give some hints about where to find the methodolagy for these? thanks.

Balmung · January 8th, 2008, 11:50 am

Search for "CDS JPMORGAN" and you'll find a description (document) of the first approach.Regards

Balmung · January 8th, 2008, 11:51 am

... in BB.

gkmitov · January 8th, 2008, 8:19 pm

QuoteOriginally posted by: seanmokHi everybody,I have seen four models used in Bloomberg (page: CDSW) for CDS pricing:1. JP Morgan2. Bloomberg3. discount spread4. Modified Hull-WhiteCan you please give some hints about where to find the methodolagy for these? thanks.Here is the JP Morgan model

seanmok · January 9th, 2008, 9:03 pm

Thanks. Actually, I have found this article, just not sure if this is the one. However, when I tried this model in Bloomberg(CDSW) with downward shape credit spread curve, it turned out to be "Error calculation". Anybody has any idea why?QuoteOriginally posted by: gkmitovQuoteOriginally posted by: seanmokHi everybody,I have seen four models used in Bloomberg (page: CDSW) for CDS pricing:1. JP Morgan2. Bloomberg3. discount spread4. Modified Hull-WhiteCan you please give some hints about where to find the methodolagy for these? thanks.Here is the JP Morgan model

Wibble · January 10th, 2008, 10:43 am

you can't have negative hazard rates so there's a maximum decrease at each step

seanmok · January 10th, 2008, 10:29 pm

QuoteOriginally posted by: Wibbleyou can't have negative hazard rates so there's a maximum decrease at each stepthanks, I feel the same way, which is about the steepness of downward shape spread curve. Can you please explain more or let me have some paper explaining about hazard rate?

NoelWatson · January 11th, 2008, 6:35 am

QuoteOriginally posted by: seanmokQuoteOriginally posted by: Wibbleyou can't have negative hazard rates so there's a maximum decrease at each stepthanks, I feel the same way, which is about the steepness of downward shape spread curve. Can you please explain more or let me have some paper explaining about hazard rate?No arbitragehttp://www.noelwatson.com/blog/PermaLink,guid, ... 0405f.aspx

Wibble · January 11th, 2008, 6:55 am

have a look at the Hull and white paper 'valuing credit default swaps I' (2000) on defaultrisk.com. It was one of the first papers to introduce hazard rates and has a simple worked example. Best way to get a feel for this is to build a spreadsheet and play around with the spreads and hazard rates

Alii · January 17th, 2008, 8:20 am

re: CDS_JPM1.pdfYeah oddly that is NOT exactly how the JPM model is implemented in Bloomberg. I have spent far too much time (hours that I will never see again!) replicating Bloomberg's figures. It uses a step function hazard rate, i.e. exponentially decreasing survival probabilities rather the linear interpolation on probability suggested here. Integration appears to be performed exactly i.e. analytically.

seppar · January 17th, 2008, 4:41 pm

QuoteOriginally posted by: Aliire: CDS_JPM1.pdfYeah oddly that is NOT exactly how the JPM model is implemented in Bloomberg. I have spent far too much time (hours that I will never see again!) replicating Bloomberg's figures. It uses a step function hazard rate, i.e. exponentially decreasing survival probabilities rather the linear interpolation on probability suggested here. Integration appears to be performed exactly i.e. analytically.Alii is right - replicating BB figures can be a daunting exercise. There are infinetely many ways of discretisizing the premium and default legs of the CDS. Typically one uses something similar to JP paper. The hazard rate is assumed to be piece-wise constant - you back out lambda1 by computing the default probability up to the first quoted spread maturity and matching this spread, then lambda1 is fixed and lambda2 is backed out from the second spread ...

Anssna · February 25th, 2009, 4:36 pm

Just learned from someone (Kun Zu) in LinkedIn group "Quant Finance" that "J.P. Morgans CDS Analytical Engine Available as Open Source".http://www.isda.org/press/press012909.htmlI also implemented a C# version of this model and output/diagram in Excel. To get a smooth curve of default probabilites, cubic spline interpolation is used for those quarterly dates on which spreads don't exist.