PaulYou commented that the current stage of credit risk modelling is in disarray (of course paraphrasing) in your new book. I have in my possession the wonderfully titled book Credit Risk:Modelling Valuation and Hedgingby Tomasz R. Bielecki and Marek Rutkowski. Although I haven’t read it cover to cover in great detail, the approach taken by each modeller were/are pretty much similar. If this path of modelling is not the right approach, how will you address the issues concerning credit risk modelling today?I am reading now LossCalc™: Moody’s Model for Predicting Loss Given Default (LGD) by Greg M. Gupton of Moody's Investment Service, and Roger M. Stein of Moody's Investors Service. Any comment?Your comments are much appreciatedMabadu

Mabadu,The articles are surely very interesting but the links are down...

Credit risk seems to be playing catch-up to other areas of quantitative finance in terms of the techniques used to model it. Most of the current research I have seen in credit risk from the academic side basically constitutes marginal improvements of techniques developed in the 1970's and 80's. On the industry side, things seem to be moving even slower (at least Duffie and Jarrow are publishing some interesting things from the academic side.)I think there are a couple of reasons for this. One is the nature of the problem. Credit risk is something which by definition can not be observed until the default event, and default is a very rare event. This has led to all kinds of attempts to describe default risk in terms of latent variables, stopping times, etc, but it is very difficult to test a hypothesis about the drivers of default risk (unlike, say, interest rates.)The second is probably a structural reason. Many credit markets, like corporate bonds, lack the liquidity of interest rate and fx markets. It is difficult to model default risk when a large part of what you observe in the market is the result of liquidity or technical factors. All the models I have seen that attempt to model default risk analytically assume that liquidity is either negligible or easily modeled. Unfortunately, this is not the case in many markets.One promising thing is that the field is still fairly wide open, and there is a lot of capital with exposure to credit risk, so I expect some interesting research to come from this area in the future.

Duffie/Singleton on Credit Risk Modeling for Financial Institutions,October 13 - 18, 2002 (at Stanford)CREDIT 2002 Conference - Assessing the Risk of Corporate Default,September 19-20, 2002, Venice, Italy

I think there are a couple of reasons for this. One is the nature of the problem. Credit risk is something which by definition can not be observed until the default event, and default is a very rare event. >>For low-graded bonds, the probability of default is not too low - it's about 20 % in a year. For investment-rated bond, the default probability is indeed very low. My feeling is that the extreme value theory may be of use here, but I don't know of any paper that would take this approach to credit risk modelling. All the models I have seen that attempt to model default risk analytically assume that liquidity is either negligible or easily modeled. Unfortunately, this is not the case in many markets. >>Check out recent papers by Jarrow on liquidity risk modelling.

Scholar, I agree with you that Jarrow is one of the people in the field of credit risk who is doing interesting new work. Do you know where I can get a copy of the paper you mentioned?As for the probability of default of low rated bonds being 20%, that might be true in the distressed market, but even the securities that are deep in high yield territory default much less frequently than 20% a year. I think part of the problem is that the lower credit stuff, which defaults frequently enough to actually have a decent no. of observations, is also where liquidity becomes very bad and dominates pricing.Default is very rare in investment grade (at least it used to be), and in the securitized world it happens even less frequently.

jaiburu stork, here are two of jarrow's papers on liquidity:Janosi, T., Jarrow, R. and Yildirim, Y., (2001), “Estimating Expected Losses and Liquidity Discounts Implicit Debt Prices”, Cornell University Jarrow, R. and Subrahmanyam, A., (1997), “Mopping up Liquidity”, Risk, December 1997, pp. 170-173i think the first is online, the second you will need to get from risk. the following is one i found useful:Ericsson, J. and Renault, O., (2000), “Liquidity and Credit Risk”, McGill University, Canada / Université Catholique de Louvain, Belgium